Re overy of 1933

∗

Margaret M. Ja obson

†Eri M. Leeper

‡Bru e Preston

�

Abstra t

When Roosevelt abandoned the gold standard in April 1933, he onverted what had

been e�e tively real government debt into nominal government debt and opened the

door to implementing an unba ked �s al expansion. We argue that he followed a state-

ontingent �s al rule that ran nominal debt-�nan ed primary de� its until the pri e

level rose and e onomi a tivity re overed. Theory suggests that government spending

multipliers an be substantially larger when �s al expansions are unba ked than when

they are tax-ba ked. VAR estimates suggest that primary de� its made quantitatively

important ontributions to raising both the pri e level and real GNP from 1933 through

1937. The eviden e does not support the onventional monetary explanation that gold

revaluation and gold in�ows, whi h were permitted to raise the monetary base, drove

the re overy independently of �s al a tions.

Keywords: Great Depression; monetary-�s al intera tions; monetary poli y; �s al pol-

i y; government debt

JEL Codes: E31, E52, E62, E63, N12

∗De ember 3, 2018. We would like to thank Nobu Kiyotaki, Tom Sargent, Ellis Tallman, Andrew S ott,

and parti ularly Mike Bordo, Tom Coleman, Chris Sims, and Tao Zha for useful omments. We also thank

George Hall and Tom Sargent for providing their government debt data. Any opinions, �ndings, and on lu-

sions or re ommendations in this material are those of the authors and do not ne essarily re�e t the views

of the National S ien e Foundation. Ja obson a knowledges support from the National S ien e Founda-

tion Graduate Resear h Fellowship Program under Grant No. 2015174787. Preston a knowledges resear h

support from the Australian Resear h Coun il, under the grant FT130101599.

†Indiana University; marmja o�indiana.edu.

‡University of Virginia and NBER; eleeper�virginia.edu

�

University of Melbourne; bru e.preston�unimelb.edu.au.

1 Introdu tion

Franklin D. Roosevelt's monetary and �s al poli ies pulled the United States out of the

Great Depression. His �rst step was monetary: Ameri a redu ed the gold ontent of the

dollar, abandoned the promise to onvert dollars to gold, and abrogated the gold lause on

all urrent, past, and future ontra ts. This paper emphasizes his se ond, �s al, step: his

administration expanded government spending, �nan ed that spending with nominal bonds,

and dissuaded people from believing that the bonds would be fully ba ked by future taxes.

Be ause the monetary omponents�devaluing the dollar and revoking onvertibility�were

ne essary for the �s al step to work, this narrative is about joint monetary-�s al a tions.

When Roosevelt shu ked o� the gold standard's straightja ket, he was freed to exploit

the nominal nature of government debt. If dollars are onvertible to gold, even dollar-

denominated government liabilities are real obligations. Credibility of the gold standard

rested on government standing ready to raise the real taxes to a quire the requisite gold

[Bordo and Kydland (1995)℄. By revoking onvertibility, Roosevelt enhan ed his poli y

options. He ould de ide to ontinue the orthodox poli y that new debt begets new taxes or

to break from the past and allow pri es to revalue outstanding bonds. Early in his presiden y,

Roosevelt hose the latter option.

Our thesis hallenges the onventional wisdom that re overy had little to do with �s al

poli y. S holars from Brown (1956) to Romer (1992) to Fishba k (2010) maintain that �s al

de� its during Roosevelt's �rst term were too small to lose the gaping gap in output.

1

Those

e onomists base their on lusion on a narrowly onstrued �s al transmission me hanism. The

government raises real spending, dire tly in reasing real aggregate demand. Higher demand

propagates through higher real expenditures and in ome, eventually to raise output by a

multiple of the initial �s al expansion. We all this me hanism �Keynesian hydrauli s,� to

use Coddington's (1976) evo ative label.

Nominal debt doubled before the end of Roosevelt's se ond term. Under Keynesian hy-

drauli s, the resulting expansion in nominal demand provides no additional e onomi stim-

ulus. Brown (1956) and the studies that followed expli itly ex lude government borrowing

from their analyses. Keynesian hydrauli s impli itly assumes that higher taxes extinguish

all wealth e�e ts from higher nominal debt. That assumption e�e tively ontinues to treat

government debt as a real obligation, denying that the suspension of gold onvertibility fun-

damentally altered the nature of government debt and the �s al options available to poli y

makers after 1933.

We broaden the perspe tive on �s al transmission to in lude both Keynesian hydrauli s

and a vehi le by whi h government debt dynami s a�e t e onomi a tivity. When nominal

government debt expands without raising expe ted taxes, private-se tor wealth and aggre-

gate demand in rease via a onventional Pigou-Keyne-Patinkin e�e t. Roosevelt exer ised

this option��unba ked �s al expansion��to implement a state- ontingent poli y: run debt-

�nan ed �s al de� its until the Ameri an e onomy re overs.

Our perspe tive omplements and elaborates Ei hengreen's (2000) on lusion that �. . . the

fundamental hange in poli y making in the 1930s was not the Keynesian revolution, but

the `nominal revolution'�the abandonment of the gold standard for managed money.� To

1

See also Chandler (1971), Peppers (1973), Beard and M Millin (1991), Raynold, M Millin, and Beard

(1991), Ei hengreen (2000), and Steindl (2004).

1

Ja obson, Leeper, & Preston: 1933

rea h our perspe tive, de�ne �money� as �nominal government liabilities.� Nothing ompels

poli y makers to ba k expansions in either omponent of nominal liabilities�base money

or bonds� with higher taxes. When they don't, debt-�nan ed �s al expansion be omes a

potent poli y tool.

1.1 The Poli y Problem

By the time Roosevelt was sworn in as the 32

nd

president of the United States in Mar h

1933, the e onomy had been de lining for over three years. Relative to the third quarter

of 1929, real GNP was 36 per ent lower while urrent-dollar GNP was 57 per ent smaller;

industrial produ tion had fallen by half; unemployment had in reased 22 per entage points;

and government debt had grown from 16 per ent to over 40 per ent of output. Although

his �rst a ts salvaged a banking system left reeling by three onse utive rises, Roosevelt's

fo us never strayed far from those ma roe onomi fa ts.

One fa t �gured prominently in his thinking: the pre ipitous de line in overall pri es

bankrupted the farmers and homeowners who had in urred nominal debts at elevated pri e

levels. Those itizens were also among Roosevelt's strongest supporters. Figure 1 en ap-

sulates the poli y problem. FDR felt that the key to e onomi re overy lay in returning

overall pri es to their 1920s levels, to a hieve �. . . the kind of a dollar whi h a generation

hen e will have the same pur hasing power and debt-paying power as the dollar we hope to

attain in the near future� [Roosevelt (1933 )℄. The problem was that in the 1920s the pri e

level was 60 per ent above the long-run average to whi h it had to revert to maintain gold

onvertibility at the parity that prevailed over the previous entury.

1840 1850 1860 1870 1880 1890 1900 1910 1920 1930

80

100

120

140

160

180

Mean 1921-1929 = 159.9

Mean 1834-1933 = 100.0

Figure 1: Consumer pri e index sin e the 1834 Coinage A t set the pri e of one oun e of gold

at $20.67. Res aled to make mean from 1834�1933=100. Sour e: O� er and Williamson

(2018) and authors' al ulations.

Roosevelt's obje tive to return the pri e level permanently to that high level was in on-

sistent with remaining on the gold standard at the histori al onversion rate. FDR pur-

2

Ja obson, Leeper, & Preston: 1933

sued a triple-barreled approa h to the problem. The exe utive bran h�with Congressional

approval�took ontrol of monetary poli y from a Federal Reserve that by all a ounts had

been �inept� sin e the depression started.

2

The monetary omponent sharply redu ed the

gold ontent of the dollar; it then evolved into omplete abandonment of the gold standard

and abrogation of gold lauses on all publi and private ontra ts.

The se ond barrel ran �emergen y� �s al de� its �nan ed by new issuan es of nominal

Treasury bonds. Emergen y spending served two purposes. It provided mu h-needed relief

through a vast array of works programs. But the modi�er �emergen y� also ommuni ated

the temporary and state- ontingent features of the �s al program.

Politi al strategy, whi h was ru ial to establish the unpre edented �s al program was

redible, omposed the third barrel. Roosevelt made re overy the poli y priority; higher,

for example, than the last entury's �s al orthodoxy. The president found innovative ways

to persuade the people the stakes of re overy were unpre edentedly high. On the domesti

front, he feared �agrarian revolution� and �amorphous resentment� of e onomi institutions.

3

Internationally, FDR onjured images of European fas ism. In advisor George F. Warren's

words, Roosevelt fa ed �a hoi e between a rise in pri e or a rise in di tators.�

4

The president

framed e onomi re overy as �a war for the survival of demo ra y� [Roosevelt (1936a)℄.

5

Jalil

and Rua (2017) present eviden e that in the se ond quarter of 1933 in�ation expe tations

pi ked up rapidly. That eviden e suggests the third barrel su eeded to onvin e people that

Roosevelt would experiment with selling bonds that do not portend higher taxes, at least

temporarily.

1.2 What We Do

The paper pla es FDR's poli y a tions in the politi al and intelle tual ontext of the times.

That ontext drives the narrative. Desperate times an engender reative measures. Despite

running for o� e on his belief in sound �nan e, Roosevelt was at root a pragmatist, willing

to experiment with the e onomi levers at his disposal�and even some levers that were not.

Several theoreti al results underpin our narrative:

1. Under a lassi al gold standard with �xed parity, monetary and �s al poli ies are not

free to a hieve any desired pri e level.

2. Unba ked �s al expansion is infeasible under a lassi al gold standard.

2

Friedman and S hwartz (1963, p. 407) hara terize their adje tive �inept� for monetary poli y as a �plain

des ription of fa t.� Also see Wi ker (1965) and Meltzer (2003) for similar assessments.

3

In O tober 1933, FDR told a group of �nan ial advisors that the gold-buying poli y the Administration

pursued averted �an agrarian revolution in this ountry� Blum (1959, p. 72). Leu htenburg's (1963) aptly-

titled hapter, �Winter of Despair,� do uments that by the winter of 1932�33, e onomi despair transformed

into �amorphous resentment� of the e onomi institutions that people blamed for the depression.

4

This quotation is found in Rau hway (2014, p. 4) and Rau hway (2015, h. 5), who lays out Warren's

in�uen e in ontext. See also Sumner (2001).

5

As early as February 1933, Marriner E les, in his apa ity as a private banker, testi�ed to the Senate

Finan e Committee that in the absen e of federal government intervention into the e onomy, �we an only

expe t to sink deeper in our dilemma and distress, with possible revolution, with so ial disintegration, with

the world in ruins, the network of its �nan ial obligations in shreds, with the very basis of law and order

shattered� [E les (1933, p. 705)℄.

3

Ja obson, Leeper, & Preston: 1933

3. Unba ked �s al expansion permanently raises the pri e level.

4. Government spending and transfer impa ts from unba ked �s al expansion generally

ex eed those from tax-ba ked �s al expansion.

We bring both informal and formal empiri al eviden e to bear on the thesis. Surprise

in�ation signi� antly redu ed the value of government debt. Over the seven years after

Ameri a left the gold standard, nominal debt rose 30 per ent more than real debt. Negative

real returns on the government bond portfolio�both a tual and surprise�be ame more

prevalent in that period. Government debt, whi h was 16.4 per ent of GNP in the last

quarter of 1929, rose to 42.3 per ent by the �rst quarter of 1933. Although nominal debt

doubled over the next de ade, it averaged only 41.6 per ent of GNP to belie the riti s'

hysteria about �s al sustainability.

Identi�ed VAR eviden e �nds that temporary �s al expansions produ e persistent in-

reases in output, the pri e level, the monetary base, the market value of nominal govern-

ment debt, and the monetary gold sto k. Fis al disturban es are also important sour es

of �u tuations in those variables and a ount for signi� ant fra tions of the k-step-aheadfore asts errors in real GNP and the pri e level. Although the VAR re overs the patterns

of orrelation that underlie onventional monetary explanations of the re overy, the VAR

points to �s al, rather than monetary or gold, sho ks as the genesis of those omovements.

2 Politi al and Intelle tual Context

Roosevelt's de ision to leave the gold standard and re�ate arose against a ba kdrop of a

growing politi al and intelle tual onsensus that higher retail and wholesale pri es were

riti al to re overy of wages, employment, investment, and onsumption. The banking risis

of February�Mar h 1933 heightened expe tations of a dollar devaluation as politi al pressure

mounted against maintaining the gold standard at the existing parity.

6

To avoid apital losses

from the banking pani , foreign depositors in U.S. banks liquidated their dollar balan es and

onverted them to gold, pushing gold reserves lose to their statutory minimums, parti ularly

at the New York Fed. The bank would have had to raise its dis ount rate in the middle of

a banking pani to attra t gold from abroad to re tify dwindling gold reserves. To avoid

further strain on the beleaguered �nan ial se tor, Senator Elmer Thomas advo ated issuing

unba ked urren y to raise the pri e level to its 1920s level and Senator Tom Connally

proposed redu ing the gold ontent of the dollar by one-third. Finan ial and politi al for es

were aligning against the gold standard.

Those realignments were e hoed by a amp of e onomists who agitated for re�ation.

Irving Fisher's (1932; 1933b) debt-de�ation theory argued that when the private se tor is

over-indebted, a falling pri e level triggers a sequen e of events�lower asset pri es, higher

real interest rates, ontra tion of bank deposits, de rease in pro�ts, redu tion in output,

rising unemployment, bank runs, and so on�that drivesdriving the e onomy into depression.

Viewing nominal in ome through the equation of ex hange, Fisher advo ated government

poli ies designed to raise the money supply and velo ity.

Fisher arried on extensive orresponden e with the president and met with him several

times to dis uss his e onomi proposals. In an April 30, 1933 letter to Roosevelt, Fisher

6

This exposition draws on Ei hengreen (1992), parti ularly hapter 11.

4

Ja obson, Leeper, & Preston: 1933

(1933a) wrote, �No one is happier than I over the prospe t of the passage of the re�ation leg-

islation,� referring to the Agri ultural Adjustment A t, whi h in luded the Thomas Amend-

ment giving the president unpre edented powers to re�ate. George F. Warren, though, had

the ear of the president. Pearson, Meyers, and Gans (1957, p. 5598), a detailed des ription

of Warren's role in Roosevelt's inner ir le, begins with the unequivo al, �George F. Warren

was the �rst person who ever advised a President of the United States to raise the pri e of

gold.�

Keynes (1933) wrote an open letter to Roosevelt, published in the New York Times,

alling for the U.S. government �. . . to reate additional urrent in omes through the expen-

ditures of borrowed or printed money.� Although today Keynesian stimulus often is narrowly

onstrued as the real me hanisms of Keynesian hydrauli s, Keynes's emphasis in this letter

is on �governmental loan expenditure� as �the only sure means of obtaining qui kly a rising

output at rising pri es.� Keynes pres ribed unba ked �s al expansion: nominal debt-�nan ed

de� its with no promise to raise future taxes to pay o� the debt.

We do not laim that Roosevelt ons iously engineered an unba ked �s al expansion.

Nor do we believe that he had in mind the pre ise e onomi me hanisms that we identify as

the sour e of the re overy. There were false starts, su h as the National Industrial Re overy

A t of 1933, whi h in addition to being ruled to ontain un onstitutional features, likely

slowed re overy [Cole and Ohanian (2004)℄. But his �try anything� ma roe onomi approa h

ontained the essential ingredients for an unba ked �s al expansion: suspension of the gold

standard, a ommitment to run debt-�nan ed emergen y de� its until spe i�ed parts of the

state of the e onomy improved, and a poli y de ision not to sterilize gold in�ows, whi h

permitted the monetary base to grow without further in reases in government indebtedness

for monetary reasons.

The paper does not try to use a formal model to reprodu e re overy-period data, as

Cole and Ohanian (2004) and Eggertsson (2008) do. In that tumultuous period, e onomi

agents onfronted an entirely new and still-evolving e onomi stru ture. Interpretations that

rely on modeling onventions like well-understood poli y rules and rational expe tations are

di� ult to align with the histori al fa ts. Instead, we use theory to frame the issues to to

inform how we interpret the history and the data.

3 Conta ts with Literature

Our argument that the joint monetary-�s al mix that underlies an unba ked �s al expansion

was the sour e of the re overy in the 1930s ontrasts with existing explanations whi h fre-

quently attribute diminished roles to both monetary and �s al poli y. Existing studies argue

that the ombination of dollar devaluation, the departure from the gold standard, regime

hange, expansion of the monetary base, and rising in�ation expe tations a ount for the

re overy. Our unba ked �s al expansion interpretation broadly agrees with many of these

arguments, but links them to the monetary and �s al poli ies of the 1930s.

Another distin tion on erns the view that monetary poli y made no substantive ontri-

bution to the re overy. Friedman and S hwartz (1963), for example, on lude the immediate

re overy �owed nothing to monetary expansion� [p. 433℄. Wi ker (1965) attributes Fed ina -

tion to a leadership va uum and the Fed's in omplete understanding of how monetary poli y

a�e ts the e onomy and the pri e level. Meltzer (2003, p. 273) �atly de lares that �. . . in

5

Ja obson, Leeper, & Preston: 1933

the middle and late thirties, just as in the early thirties, the Federal Reserve did next to

nothing to foster re overy.�

We argue that by pegging short-term interest rates throughout the 1930s, the Fed per-

mitted unba ked �s al expansion to re�ate the e onomy. Expansions in nominal debt that do

not portend higher future taxes raise household wealth at prevailing pri es and interest rates.

Bond holders onvert higher wealth into higher aggregate demand. Some of the in reased

demand shows up in aggregate pri e levels, but if pri es do not adjust instantaneously, some

demand raises real e onomi a tivity. By pegging interest rates, monetary poli y prevents

the nominal debt expansion from raising debt servi e enough to put debt on an explosive

path. Federal Reserve poli y performed the riti al role of stabilizing government debt.

Pegged rates also do not �ght against the higher pri e levels needed to bring the real market

value of debt in line with the expe ted present value of the primary surpluses that ba k

debt.

7

Monetary and �s al poli y are equal partners in su essful unba ked �s al expansion.

The e onomi onsequen es of the unba ked �s al expansion that began in 1933 ra-

tionalize why on erns that expanding federal debt would threaten the U.S. government's

reditworthiness were not realized. Studenski and Krooss (1952, p.428) summarize a key

feature of unba ked �s al expansion:

�In its early years, the New Deal administration itself believed that the publi

redit ould not sustain ontinuous budgetary de� its and in reases in the publi

debt. But in pra ti e this also proved in orre t. The publi redit did not ollapse

under the burden of in reased publi debt. On the ontrary, government redit

grew stronger, interest rates on new government borrowing de lined steadily, and

the Treasury found it in reasingly easy to �nan e its operations.�

Unba ked expansions raise pri es and real GNP to ensure that higher nominal debt does not

transform into a higher debt-output ratio.

The initial impetus for re overy ame from dollar devaluation and departure from the

gold standard, whi h signaled a hange in poli y regime that raised in�ation expe tations,

a ording to the onsensus view. We agree that these elements all ontributed to the re ov-

ery, but argue they annot a ount for the rapid pi k up in the pri e level and output in

isolation. Temin and Wigmore (1990) o�er eviden e that dollar devaluation in 1933 signaled

that Roosevelt had abandoned the de�ation asso iated with adheren e to the gold standard

and that the lower dollar dire tly in reased aggregate demand and indire tly raised pri es

and produ tion throughout the e onomy. Hausman (2013) provides eviden e of Temin and

Wigmore's hypothesis by showing that in reased agri ultural in omes bolstered auto sales

in rural areas. Romer (1992), however, makes a for eful ase that the dollar depre iation

following the departure from the gold standard in late April 1933 annot a ount for the

sustained in rease in in�ation in subsequent years. We agree with Romer and point out�as

do Jalil and Rua (2017)�that both Britain and Fran e experien ed similar depre iations

in their urren ies upon exit from gold, yet pri es and output did not rise as in the United

States.

7

This me hanism is des ribed in detail in a growing literature that began with Leeper (1991), Woodford

(2001), Sims (1994), and Co hrane (1999).

6

Ja obson, Leeper, & Preston: 1933

Our work omplements Jalil and Rua's narrative eviden e on the role of rising in�ation

expe tations in the re overy of 1933. We ground those expe tations in the monetary-�s al

poli y mix.

The argument di�ers from Eggertsson (2008), who emphasizes a regime hange in poli y

dogmas from Hoover to Roosevelt and relies on new Keynesian me hanisms for es aping from

the lower bound on the nominal interest rate, with expe tations an hored on an eventual

return to the onventional a tive monetary/passive �s al poli y mix.

8

Eggertsson's story

rests on the oordinated a tion of monetary and �s al poli y to maximize household utility.

In the presen e of distortionary taxation, higher de� its provide an in entive for the Fed to

keep interest rates low for an extended period of time, to manage the value of outstanding

debt. Monetary poli y mitigates the distortions of tax poli y by ommitting to generate

in�ation when the Fed has the freedom to do so�that is, on e the zero lower bound eases

to bind. In this way, the time- onsistent poli y generates the same stimulatory me hanisms

that Eggertsson and Woodford's (2003) optimal ommitment poli y delivers.

This interpretation fa es several di� ulties. Does eviden e support the degree of poli y

oordination that Eggertsson's model requires? E les (1951) des ribes a highly de entralized

Federal Reserve, both in its operations and in its obje tives, an a ount that Wi ker (1966),

Meltzer (2003), and Wheelo k (1991) on�rm. Federal Reserve o� ials frequently voi ed

on erns about the prospe t of in�ation, even during the de�ationary years in the early 1930s

[Meltzer (2003, p. 280)℄. The volume of those voi es rose in FDR's �rst term in response

to �imprudent� �s al poli ies [ itation℄. Eggertsson's me hanism leans heavily on rational

expe tations at a time when the entire monetary system had no pre edent. It is di� ult

to square that history with Eggertsson's sophisti ated and single-mindedly in�ationary Fed

behavior.

History was not nearly as linear as our unba ked �s al expansion interpretation makes it

seem. Disparate viewpoints about the depression battled for �the soul of FDR,� in Stein's

(1996, h. 6) memorable phrase. A 1932 �Memorandum� written by three young Harvard

e onomists ni ely distills those disparate views. The do ument denoun es �the failure on the

part of the government to adopt other than palliative measures� to ombat the depression

[Currie, White, and Ellsworth (2002, p. 534)℄. Viewpoints Roosevelt ontended with in-

luded: (1) e onomists who believe the depression annot be stopped and any e�orts to do

so interfere with the �natural� fun tions of the e onomy; (2) those who believe the e onomy

is so poorly understood that government e�orts are likely to make matters worse; (3) some

who adopt the view that depressions are leansing and purge ine� ien ies; (4) a group, like

the Memorandum's authors, who �believe that re overy an and should be hastened thru

[si ℄ adoption of proper measures.�

9

Roosevelt learly sided with the fourth group, at least in the early years of the re overy.

8

Leeper (1991) de�nes an a tive poli y authority as free to pursue its obje tive, while a passive authority

is onstrained by the behavior of the a tive authority and optimizing private behavior. In onventional

models, a determinate bounded rational expe tations model requires either an a tive monetary poli y with

a passive �s al poli y or vi e versa.

9

Two authors went on to play riti al roles in poli y: Currie at the Federal Reserve Board, Treasury

and the White House; White at the Treasury where, together with Keynes, he reated the Bretton Woods

system.

7

Ja obson, Leeper, & Preston: 1933

4 Why Unba ked Fis al Expansion?

Contemporary supporters and riti s understood that Roosevelt's pri e-level obje tive en-

tailed a permanent in rease in pri es to 60 per ent above their long-run average. But a

permanent revaluation of the dollar pri e of gold required leaving the gold standard.

Result 1. Under the gold standard with a �xed parity�the lassi al gold standard�monetary

and �s al poli ies annot a hieve any desired pri e level.

Straightforward e onomi logi underlies this result.

10

Private holdings of gold, whi h

standard asset-pri ing reasoning determines, establish the goods value of gold�the aggregate

pri e level. The Euler equation for private gold demand implies that

P gt

Pt

= Et

∞∑

T=t

qt,TuG,T

uc,T

(1)

where P gt is the dollar pri e of gold, Pt is the pri e level, qt,T is the sto hasti dis ount fa tor,

uG,T is the marginal utility of gold holdings, and uc,T is the marginal utility of onsumption.

When the dollar pri e of gold is �xed at P gt = P g

, expression (1) implies that the marginal

rate of substitution between gold and onsumption uniquely determines the equilibrium pri e

level.

Monetary poli y must passively adjust to a ommodate the pri e level onsistent with the

pegged pri e of gold. Fis al poli y must passively adjust primary surpluses to provide gold

ba king for outstanding government debt at that pri e level. This establishes that leaving

the gold standard and abandoning onvertibility were ne essary to a hieve FDR's pri e-level

obje tive.

De�nition 2. Unba ked �s al expansion in reases government expenditures on pur hases

and transfers, issues nominal bonds to over the de� it, and persuades people that surpluses

will not rise to �nan e the bonds.

Simple theory makes this de�nition pre ise and illustrates the pri e-level onsequen es

of unba ked �s al expansion. A representative household re eives a onstant endowment,

derives utility from onsumption and real money balan es, and hold initial nominal wealth in

the form of nominal money and bonds, A0 ≡ M−1+B−1. Nominal bonds sold at t sell at pri e1/(1 + it) and money earns no interest. The household's intertemporal budget onstraint at

time 0 is

E0

∞∑

t=0

q0,t

[

ct +it − 1

itmt

]

=A0

P0

+ E0

∞∑

t=0

q0,t [yt − τt] (2)

q0,t is the sto hasti dis ount fa tor for the date-0 value of goods at t, mt is real money

balan es, and τt is lump-sum taxes net of transfers. Money demand yields the liquidity

preferen e s hedule mt = L(it, ct).To lose the model, we assume the entral bank pegs the nominal interest rate, it = i,

as the Federal Reserve did after 1933. Fis al poli y sets τt = τ + εt, where Etεt+j = 0 for

10

See Barro (1979) or Goodfriend (1988) for details.

8

Ja obson, Leeper, & Preston: 1933

j > 0, and government pur hases are zero. Applying these poli y rules, imposing goods- and

bond-market learing on (2), and evaluating expe tations yields the equilibrium ondition

M−1 +B−1

P0

= L(i, y) + τ0 +β

1− βτ (3)

The real value of government liabilities equals the expe ted present value of seigniorage

revenues plus primary surpluses.

Lower τ0 is an unba ked �s al expansion. Higher transfers with no o�setting future

taxes shift resour es from the government to households. This positive wealth e�e t indu es

households to attempt to raise their onsumption paths. Higher demand for goods raises

their pri e, P0, whi h redu es the real value of the household's nominal assets, A0/P0. This

negative wealth e�e t must be large enough to eliminate the ex ess demand for goods at

time 0, and make households happy to onsume their endowments.

Corollary 3. Unba ked �s al expansion is infeasible under a lassi al gold standard.

Unba ked �s al expansion requires a tive �s al behavior; the government does not use

future surpluses to stabilize debt. Condition (3) uniquely determines the pri e level as a

fun tion of the expe ted present value of primary surpluses in luding seigniorage revenues�

the right side�and outstanding nominal government liabilities. Asset-pri ing ondition (1)

determines the pri e level as a fun tion of the gold pri e, P g, and prevailing onditions in

the gold market. These two pri e levels will generally be di�erent.

When the pri e level onsistent with P gis too low to satisfy (3), the real value of debt

ex eeds its real ba king. Households will over-a umulate government bonds to violate their

optimality onditions. When the pri e level under the gold standard is too high, households

will refuse to buy bonds, and the government will violate its budget onstraint. By either

out ome, no equilibrium exists.

Result 4. Unba ked �s al expansion permanently raises the pri e level.

A one-time unba ked �s al expansion raises P0 in equilibrium ondition (3). To see that

this in rease is permanent, examine how nominal government liabilities at time 0 hange.

Both real money balan es, M0/P0 = L(i, y), and real debt, B0/P0 = τ /(1 − β), remain

un hanged be ause they do not depend on τ0. With the hange in pri e level, ∆P0, given by

the equilibrium ondition, both M0 and B0 expand in proportion to ∆P0. In the absen e of

any further disturban es, nominal liabilities remain at those permanently higher levels, as

does the pri e level.

11

These theoreti al points establish that an appropriately s aled unba ked �s al expansion

ould, in prin iple, a hieve FDR's pri e-level obje tive and that ending onvertibility of

dollars for gold was a ne essary �rst step. But why did Roosevelt opt for a �s al, rather

than a monetary, solution?

11

Be ause the expansion in M0 depends on L(i, y), this is not onventional money �nan ing of de� its, as

in Sargent and Walla e (1981). Instead, the money supply expands passively to ensure the money market

ontinues to lear at the pegged nominal interest rate i.

9

Ja obson, Leeper, & Preston: 1933

4.1 Monetary Poli y

In the wake of the Federal Reserve's �ina tivity� in the worst years of the depression, Congress

feared that any re overy would be stymied by ontinued Fed ina tion.

12

The Thomas Amend-

ment of May 1933 granted the Exe utive unpre edented monetary powers, whi h in luded

�xing the gold value of the dollar, issuing greenba ks, and ordering the Fed to buy Treasury

se urities. This was a �rst step to ensure the Fed would not a t to thwart the stimulative

impa ts of �s al expansion.

Enter Klüh and Stella's (2018) argument that the Gold Reserve A t of 1934 undermined

the Fed's ability to reverse the stimulus through open-market operations. The A t gave to the

Treasury legal title to all monetary gold. Treasury bought gold by issuing gold erti� ates,

whi h ould be held only by the Fed and were redeemable in dollars only at the Treasury's

dis retion. Treasury gold pur hases raised the Fed's monetary liabilities�new Treasury

deposits at the Fed�without ommensurate in reases in liquid assets. By the end of 1936,

the Fed's total monetary liabilities were $10.89 billion, of whi h only $2.43 billion were liquid.

Over 80 per ent of the Fed's monetary liabilities were irredeemable gold erti� ates.

13

Klüh and Stella (2018, p. 4) observe that Fed o� ials �understood they ould not win a

war of attrition with the Treasury.� The Treasury ould undertake gold pur hases to expand

reserves without limit, se ure in the knowledge that it was infeasible for the Fed to sterilize

them.

Operational fa tors ombined with institutional features of the Federal Reserve in the

early 1930s to redu e the Fed to �impoten e,� a ording to E les (1951). At the time, there

was no single Federal Reserve poli y; there was a poli y for ea h regional Reserve Bank

and the Board of Governors. E les emphasizes that Reserve Banks were beholden to their

dire tors, who a ted in the private interests of bankers. Before a epting the nomination to

hair the Federal Reserve Board, E les insisted on institutional reforms that onsolidated

de ision-making power in Washington, D.C. The Banking A t of 1935, among other things,

hanged the de ision-making pro ess at the Fed, whi h E les des ribes:

�. . . before a uniform de ision ould be rea hed. . . there had to be a omplete

meeting of the minds between the governors of the 12 Reserve banks and the

108 dire tors of those banks, plus the FRB in Washington. A more e�e tive way

of di�using responsibility and en ouraging inertia and inde ision ould not very

well have been devised.� E les (1951, p. 170)

While the Fed ould not sterilize the Treasury's gold pur hases, monetary poli y also

did little to advan e Roosevelt's e onomi agenda. After only minor a tions in 1933, the

Fed ondu ted no open-market operations after November 1933. This ina tivity o urred

against a ba kdrop of urrent and former Fed o� ials publi ly expressing on erns about

run-away in�ation. After leaving his position as Fed Chairman on May 10, 1933, Eugene

Meyer wrote that �. . . the mere fa t that the Administration has assumed responsibility for

de�ning our monetary poli ies and �xing our pri e goal, indi ates a subordinate role for the

12

Meltzer (2003, p. 459), but see also Friedman and S hwartz (1963) and Wi ker (1966).

13

Board of Governors of the Federal Reserve System (1937). Total monetary liabilities are Federal Reserve

and Federal Reserve Bank notes outstanding plus bank reserves; total liquid assets are gold reserves plus

U.S. Treasuries.

10

Ja obson, Leeper, & Preston: 1933

Federal Reserve System� [Meyer (1934)℄. Adolph Miller, one of the original governors of the

Federal Reserve System, who served until 1936, was vo iferous in alling for a return to gold,

fearing the dis retion that underlies a �managed urren y,� whi h he alled �human nature

money� [Miller (1936, p. 4)℄.

At a pra ti al level, it was not lear that monetary stimulus would be e�e tive. There

was no assuran e, parti ularly on the heels of sequential banking rises, that higher reserves

would lead to higher bank deposits. Nor was it ertain that higher deposits, if they were

forth oming, would result in in reased bank loans to �nan e new investment.

As it happened, banks, worried about the Federal Reserve's failure to ful�ll its lender-

of-last-resort fun tion, behaved onservatively and expanded holdings of government bonds,

rather than loans to the private se tor. From Mar h 1933 to June 1940, annual growth rates

of narrow money far outstripped those of broad money: reserves (23.1 per ent), base (12.8

per ent), M1 (7.7 per ent), and M2 (5.2 per ent). This was a very di�erent pattern from

the 1920s when M2 averaged 3.2 per ent annual growth and reserves averaged 2.8 per ent.

4.2 Fis al Poli y

Unba ked �s al expansion served several of FDR's obje tives. Given his strong support in

Congress, parti ularly from �in�ationists� like Senators Thomas and Connally, �s al poli y

was largely under the president's dire t ontrol. Federal Reserve poli y, to FDR's frustration,

was beyond his ontrol.

Fis al poli y also served politi al obje tives. By providing immediate relief to the un-

employed, farmers, and the �forgotten man,� federal expenditures tamped down domesti

unrest. Dire t relief was a highly visible indi ator that the federal government had the

ommon man's interests at heart, helping to re-establish on�den e in poli y institutions.

Finally, e onomists and politi ians alike understood that de�ation had redistributed wealth

from debtors to reditors. Re�ation, and the �s al a tions underlying it, were deliberate

e�orts to reverse that redistribution.

14

Roosevelt's attitudes toward redistribution shone

through in a letter to Se retary of the Treasury Woodin: �I wish our banking and e onomist

friends would realize the seriousness of the situation from the point of view of the debtor

lasses�i.e., 90 per ent of the human beings in this ountry�and think less from the point

of view of the 10 per ent who onstitute reditor lasses� [Roosevelt (1933a)℄.

Roosevelt walked a �ne line on �s al poli y, seeming to maintain ontradi tory positions

simultaneously. During the 1932 ampaign for president, he harshly riti ized Hoover's

de� its and took a �Pittsburgh pledge� to balan e the budget by redu ing expenditures

[Roosevelt (1932a)℄. Just six months earlier he delivered his famous spee h about �the

forgotten man at the bottom of the e onomi pyramid� [Roosevelt (1932b)℄. That spee h

hara terized the depression as a �more grave emergen y� than World War I and alled for

a restoration of the pur hasing power of farmers and rural ommunities and assistan e to

homeowners and farmers fa ing fore losure.

14

Fisher (1934, h. VI) thoughtfully dis usses how to arrive at a �just� pri e level that balan es the losses

of borrowers and reditors. E les (1933) pointed to the redistribution of wealth as a sour e of the prolonged

depression: �During the period of the depression the reditor se tions have a ted on our system like a great

su tion pump, drawing a large portion of the available in ome and deposits in payment of interest, debts,

insuran e and dividends. . . .�

11

Ja obson, Leeper, & Preston: 1933

Six days after taking o� e, Roosevelt sent to Congress a proposal to ut federal spending

by an amount equal to nearly 14 per ent of total expenditures. Cuts eliminated government

agen ies, redu ed federal worker pay, and, most riti ally in light of the politi s of the time,

shrank veterans' bene�ts by half. When the E onomy A t of 1933 was �nally signed into law,

the spending uts amounted to a little under seven per ent of expenditures, but Roosevelt

ould point to the legislation to help establish his bona �des as a �sound �nan e� man.

Just 20 days into his administration, Roosevelt drew �ne lines on �s al matters in a press

onferen e. Asked when it might be possible to balan e the budget, the president replied,

�. . . it depends entirely on how you de�ne the term, `balan e the budget� ' [Roosevelt (1933b,

p. 13)℄. His reply spawned the distin tion between �ordinary� and �emergen y� expenditures,

whi h be ame institutionalized in Treasury Reports.

15

FDR was more omfortable with de� its by 1936. In the fa e of pre ipitous de lines in tax

re eipts, he argued that �To balan e our budget in 1933 or 1934 or 1935 would have been a

rime against the Ameri an people� [Roosevelt (1936b)℄. And in response to budget dire tor

Lewis W. Douglas's argument that the only way to proje t a balan ed budget in 1936 was

to ut spending, Roosevelt replied, �No, I do not want to taper o� [spending programs℄ until

the emergen y is passed� [Rosen (2005, p. 85)℄. On the other hand, he supported tax hikes

in 1935 and 1937.

Why did FDR wa�e so on �s al poli y? Although it is possible, as Stein (1996) suggests,

that Roosevelt was tentative and un ertain about �s al stimulus, the wa�ing may have been

deliberate. His distin tion between �ordinary� and �emergen y� government expenditures was

entral to ommuni ating that unba ked �s al expansion was state- ontingent. Linking the

state- ontingent emergen y expenditures tightly to the e onomi emergen y�through both

their timing and their labels�Roosevelt drove home their temporary nature. At the same

time, by demonstrating �s ally responsible ordinary spending, he ould reassure his riti s,

parti ularly bankers, that on e the risis passes, he would balan e the budget. Roosevelt's

January 1936 budgetary address made this point expli it when he said, �. . . it is the de� it

of today whi h is making possible the surplus of tomorrow� [Roosevelt (1936 )℄.

5 Empiri al Fa ts

This se tion presents a variety of fa ts about the state of the U.S. e onomy throughout

the 1920s and 1930s fo using on orroborative eviden e that points towards interpreting

the re overy as an unba ked �s al expansion. In the �gures that follow, we ontrast the

performan e of e onomi variables during the �gold standard� (January 1920 to Mar h 1933)

to their behavior during the �unba ked �s al expansion� (April 1933 to June 1940). Data

are quarterly. Verti al bars in the �gures at April 1933 mark Ameri a's departure from the

gold standard.

15

The reply ontinued: �What we are trying to do is to have the expenditures of the Government redu ed,

or, in other words, to have the normal regular Government operations balan ed and not only balan ed, but

to have some left over to start paying the debt. On the other hand, is it fair to put into that part of the

budget expenditures that relate to keeping human beings from starving in this emergen y? I should say

probably not. . . You annot let people starve, but this starvation risis is not an annually re urring harge.

I think that is the easiest way of illustrating what we are trying to do in regard to balan ing the budget. I

think we will balan e the budget as far as the ordinary running expenses of the Government go� Roosevelt

(1933b, pp. 13�14)

12

Ja obson, Leeper, & Preston: 1933

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940

60

70

80

90

100

110

120

130

Output

Real GNP

IP

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940

70

80

90

100

110

120

130

140

150

160

Price Levels

GNP Deflator

CPI

WPI

Figure 2: Measures of real e onomi a tivity and pri e levels. All series use 1926 base year.

Verti al line marks when the United States abandoned the gold standard. Sour es: Balke and

Gordon (1986), Federal Reserve Board, BEA and BLS from NBER Ma rohistory Database.

5.1 Ma roe onomi Indi ators

The pri e level, however measured, de reased by roughly 30 per ent from the sto k market

rash in O tober 1929 to its trough in April 1933 when the United States abandoned the

gold standard (right panel �gure 2). Although onsumer and wholesale pri es and the GNP

de�ator rose through most of the 1930s, they never regained the 1920s target levels proposed

by various poli ymakers.

Like pri es, output also plunged after the sto k market rash and rebounded with the

abandonment of the gold standard. The left panel of �gure 2 shows that real GNP fell by

roughly 25 per ent from peak to trough, as measured on an annual basis. GNP hits its

trough in the �rst quarter of 1933. Industrial produ tion dropped 45 per ent from peak to

trough and, like onsumer and wholesale pri es, began a sustained re overy in April 1933.

Unlike those pri es, GDP and industrial produ tion eventually surpassed their pre-re ession

peaks later in the de ade.

The left panel of �gure 3 shows the dollar-sterling and dollar-fran ex hange rates. The

�rst verti al line marks when the United Kingdom left gold in September 1931, whi h trig-

gered a very large dollar appre iation that was reversed in April 1933. Note that sterling's

depre iation against the dollar is roughly omparable to its subsequent appre iation.

The �gure's right panel plots the level of the GNP de�ator along with two interest rates�

the ommer ial paper rate for New York and the New York Fed's dis ount rate. Although

during the gold standard period interest rates generally followed the de line in the pri e level,

there are also several distin t deviations when rates rose sharply despite a �at or de lining

pri e level. For example, in O tober 1931, on erns about gold out�ows indu ed most Federal

Reserve Banks to raise their dis ount rates after Britain left the gold standard, even though

pri es were in free fall. The Federal Reserve banks aimed to mitigate gold out�ows resulting

13

Ja obson, Leeper, & Preston: 1933

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940

3.4

3.6

3.8

4

4.2

4.4

4.6

4.8

5

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09Value of Dollar

Dollar/Sterling

Dollar/Franc (right)

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940

1

2

3

4

5

6

7

80

90

100

110

120

130

Interest Rates & Price Level

Commercial Paper Rate

NY Fed Discount Rate

GNP Deflator (right)

Figure 3: Ex hange rates, in�ation, and interest rates. Ex hange rates in dollars per foreign

urren y; in�ation is annual (quarter over four quarters prior). First verti al line marks when

the United Kingdom abandoned the gold standard; se ond line marks when the United States

abandoned the gold standard. Sour es: Federal Reserve Board (1943).

from the appre iation of the dollar vis-à-vis the pound. Meltzer (2003, p. 280) laims that

Federal Reserve poli y de isions were mostly onsistent with the Rie�er-Burgess and real

bills do trines.

16

But these interest-rate hikes were lear attempts by the Federal Reserve to

follow the gold standard's �rules of the game� [p. 273℄.

After the abandonment of the gold standard in April 1933, the Federal Reserve pegged

interest rates near zero. Meltzer (2003, p. 413) notes that the Federal Reserve made few

hanges to the market portfolio and dis ount rate from 1933 to 1941. If anything, rates

moved against the pri e level, so the Fed was ertainly not following what today we might

all a pri e-level target. This raises the theoreti al question of how the pri e level was

determined after Ameri a left the gold standard. Eggertsson (2008) laims that Fed poli y

an hored expe tations on the belief that on e monetary poli y exited the zero lower bound,

it would follow a now-standard a tive monetary/passive �s al poli y mix. These beliefs an,

in prin iple, uniquely determine the pri e level.

The top panel of �gure 4 plots the monetary base and the monetary gold sto k and

the bottom panel plots the gold over ratio. Monetary aggregates fell in the early 1930s as

�nan ial unrest lead to ontra tions in deposits and ash hoarding by the publi . Table 1

reports that total deposits in all banks fell 30 per ent between 1929 and the low point in 1932�

33. Deposits boun ed ba k to their pre-depression levels by 1937. Loans, whi h de lined over

50 per ent never regained their previous level. Bank holdings of U.S. government obligations

largely �lled the asset void left by loans, tripling between 1929 and 1937.

The large jump in gold sto k and the ratio in 1934 stem from the revaluation of gold

to $35 an oun e. Steady in rease in the two monetary measures during the unba ked �s al

16

Meltzer (2003, p. 282) elaborates that under the Rie�er-Burgess framework, poli ymakers fo used on

borrowed reserves and short-term market interest rates as key signals of bank demand.

14

Ja obson, Leeper, & Preston: 1933

expansion period re�e ts the Roosevelt Administration's de ision not to sterilize gold in�ows.

That de ision was reversed in 1937, redu ing the growth rate of the base [Irwin (2012)℄ (see

appendix D for more details on sterilization).

For a ouple of years before the gold revaluation, the over ratio was pre ariously low,

imposing a severe onstraint on the level of the monetary base. Ei hengreen (1992) re ounts

events during February and Mar h 1933 when the New York Fed was at its statutory 40

per ent minimum gold over ratio, whi h prevented it from redis ounting bills. Initially,

other reserve banks dis ounted bills on New York's behalf. By Mar h 3 the Chi ago Fed,

whi h held the bulk of the System's ex ess gold, refused to provide further assistan e to New

York for fear that it would be unable to help banks in the Chi ago distri t. These tensions,

whi h stemmed from the absen e of a oherent national monetary poli y, exa erbated the

already tenuous state of ommer ial banks and raised doubts about the redibility of the

System's ommitment to gold parity.

O� ial revaluation of gold in January 1934 in reased the over ratio sharply and it

remained lose to 0.90 for the remainder of the de ade. Gold no longer onstrained poli y

behavior as it had before April 1933, a point that is entral to the theory of unba ked �s al

expansion that se tion 4 presents.

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 19400

5

10

15

20

25Monetary Base & Gold Held by FR, Treasury, Outside

Monetary Base

Gold

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 19400.2

0.4

0.6

0.8

1Gold Cover: Ratio of Gold to Base

Figure 4: Monetary base and gold held by Federal Reserve Banks. Verti al line marks when

the United States abandoned the gold standard. Sour e: Federal Reserve Board (1943) from

NBER Ma rohistory Database.

5.2 Poli y Behavior

Many authors have noted that adheren e to the gold standard imposed severe onstraints

on monetary and �s al poli ies by fo using poli y authorities on international onsiderations

15

Ja obson, Leeper, & Preston: 1933

1929 1932-33 1937

High Low High

Annual data

In 1939 pri es, billions of dollars

GNP 85.9 61.5 87.9

Gross domesti investment 14.9 1.1 11.4

In urrent pri es, billions of dollars

GNP 103.8 55.8 90.2

Gross domesti investment 15.8 0.9 11.4

Consumption 78.8 46.3 67.1

Biannual data

All banks, billions of dollars

Total deposits 59.8 41.5 59.2

Loans 41.9 22.1 22.1

U.S. government obligations 5.5 8.2 17.0

Table 1: Sour es: Gordon (1952, p. 390) and Federal Reserve Board (1943).

at the expense of domesti onditions [see Wi ker (1966) for dis ussions of monetary poli y

onstraints℄. Ei hengreen (2000) argues that the gold standard prevented governments from

re�ating: �So long as the gold standard remained in pla e, the ommitment to defend the

entral bank's gold reserves and stabilise the gold parity was an insurmountable obsta le to

the adoption of expansionary poli ies.� Apropos of �s al poli y under the gold standard,

when taxes must ba k government debt, is Ei hengreen's statement: �De� it spending ould

not be used. . . if de� it spending ould not be �nan ed.�

Figure 5 illustrates pre isely the onstraint on monetary poli y that Ei hengreen has

in mind. Dashed lines are interest rates and the solid line is the growth rate of the gold

sto k. A shrinking gold sto k usually indu ed Federal Reserve Banks to raise interest rates

to attra t gold from abroad, whi h arrived with a lag. And when Federal Reserve Banks

lowered interest rates, gold would �ow out of the United States. But in the 1920s, as �gure

3 shows, these interest-rate movements o urred in the fa e of a steadily falling pri e level.

The Fed's a tions were designed to stabilize ex hange rates at the expense of domesti pri es.

Our interpretation of the 1930s re overy relies on a joint monetary-�s al poli y mix that

was possible only after abandoning the gold standard. The top panel of �gure 6 plots three

measures of the federal budget surplus: gross, primary, and �ordinary,� de�ned as total

re eipts less what are labeled �ordinary� expenditures. All three measures of de� its as a

share of GNP deteriorated sharply as e onomi a tivity ontra ted in the early 1930s. Falling

surpluses stemming from de lining revenues due to lower orporate and in ome tax re eipts

and rising expenditures due to in reased publi works spending.

17

Although Roosevelt touted

the evils of de� its and was more outspoken than President Herbert Hoover in his promise to

ut expenditures, until the se ond half of the de ade he did little to onvert primary de� its

to primary surpluses.

18

De� its remained sizeable until 1936, despite growing re eipts from 1934 onward [table

17

Stein (1996, p. 25), Studenski and Krooss (1952, p. 359), and Garbade (2012, p. 2).

18

Stein (1996, p. 87) notes that, at least initially, Roosevelt was able to �rise above� his belief in redu ing

expenditures to do what he onsidered ne essary whi h was in reasing spending.

16

Ja obson, Leeper, & Preston: 1933

1920 1922 1924 1926 1928 1930 1932

2

3

4

5

6

7

-20

-15

-10

-5

0

5

10

15

20

Interest Rates & Gold Flows

Commercial Paper Rate

NY Fed Discount Rate

Gold Stock Growth (right)

Figure 5: Interest rates and growth rate of monetary gold sto k. Growth rate annual (quarter

over four quarters prior). The verti al line marks when the United Kingdom abandoned the

gold standard. Sour es: Federal Reserve Board (1943).

2℄. To reassure the publi that �s al �nan es were �sound,� Roosevelt's Treasury drew a lear

line between �ordinary� and �emergen y� government expenditures. With the ex eption of

1936, when large veterans' bonuses were paid out, Roosevelt ould laim that he balan ed

the �ordinary� budget [�gure 6℄. The bottom panel of the �gure plots the primary surplus

ex luding and in luding seigniorage revenues: evidently, seigniorage did not make signi� ant

dents in the budget de� it.

1929 1930 1931 1932 1933 1934 1935 1936 1937

Total re eipts 4033 4178 3317 2121 2080 3116 3801 4116 5294

Total expenditures

(ex luding debt retirements) 3299 3440 3780 4594 4681 6745 6802 8477 8001

�Regular� 3299 3440 3780 4594 4681 2741 3148 5186 5155

�Emergen y� 0 0 0 0 0 4004 3655 3301 2847

�Regular De� it� −734 −738 463 2473 2601 −375 −653 1070 −139De� it −734 −738 463 2473 2601 3629 3001 4361 2707

Table 2: Millions of urrent dollars. �Emergen y� expenditures are variously labeled as �emer-

gen y organization expenditures,� �major expenditures due to or a�e ted by the depression,�

�re overy and relief,� or �publi works.� Designations of types of spending as �regular� or

�emergen y� hanged over time. A negative de� it is a surplus. Sour e: Department of the

Treasury (various).

Emergen y expenditures drove budget de� its. Before 1934, non-ordinary expenditures

onsisted entirely of debt retirements. From 1934 to 1939, monthly expenditures were lassi-

�ed as general or emergen y, where emergen y spending was asso iated with relief measures

under the New Deal. Annual Treasury reports retroa tively ategorize emergen y expen-

17

Ja obson, Leeper, & Preston: 1933

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940-12

-10

-8

-6

-4

-2

0

2

4Surpluses as Percent of GNP

Primary

Gross

Ordinary

Figure 6: Surpluses de�ned as total re eipts less expenditures, ordinary or total. Primary

surplus is gross surplus less net interest payments. Seigniorage is de�ned as (Mt −Mt−1)/Pt

where M is monetary base and P is the GNP de�ator. Verti al line marks when the United

States abandoned the gold standard. Sour es: Federal Reserve Board (1943) from NBER

Ma rohistory Database, and Balke and Gordon (1986). See Appendix A for more details on

the data series.

ditures only ba k to 1933 [see appendix A.2 for details℄. Figure 7 (top panel) shows that

emergen y expenditures rose dramati ally during Roosevelt's �rst year in o� e before falling

ba k to an annual average of $3.4 billion per year until the end of 1939.

Emergen y expenditures are strongly orrelated with real GNP growth and in�ation

during the unba ked �s al expansion period. Figure 7 (bottom panel) reports rolling orre-

lations between emergen y expenditures as a share of GNP and those two ma roe onomi

aggregates. Contemporaneous orrelations are omputed with a �xed rolling window of

28 quarters, beginning with the sample 1920Q1�1926Q4 and ending with the sub-period

1933Q3�1940Q2. Correlations early in the sample, therefore, re�e t the fa t that debt re-

tirement is un orrelated with in�ation and e onomi growth. But as the window moves

forward in time, emergen y expenditures in reasingly re�e t New Deal spending on relief

and those expenditures are very strongly linked to in�ation and real GNP growth.

5.3 Keynesian Hydrauli s vs. Unba ked Fis al Expansion

Result 5. Government spending and transfer impa ts from unba ked �s al expansions typi-

ally ex eed those from Keynesian hydrauli s alone.

18

Ja obson, Leeper, & Preston: 1933

1928 1930 1932 1934 1936 1938 19400

2

4

6

0

5

10

Federal Emergency Spending

Billions of $

Percent of GNP (right)

1928 1930 1932 1934 1936 1938 1940

-0.2

0

0.2

0.4

0.6

0.8

Correlations with Real GNP Growth & Inflation

GNP

Inflation

Figure 7: Emergen y expenditures are total expenditures in ex ess of ordinary expenditures.

Rolling orrelations between in�ation and real GNP growth and emergen y federal expendi-

tures as a share of GNP omputed over a seven-year window. Sour e: Authors' al ulations.

5.4 Developments in Government Debt

If FDR had intended to engineer an unba ked �s al expansion, growth in government liabil-

ities suggests he was su essful. Nominal gross debt doubled during his �rst seven in o� e.

By omparison, seven �s al years after the �nan ial risis in 2008, U.S. gross federal debt

in reased by a fa tor of 1.8.

The left panel of �gure 8 plots index numbers for nominal and real federal debt. Taken

together, the two panels highlight entral features of unba ked �s al expansions: despite

in reases in nominal debt, real debt rises less dramati ally and there may be no in rease at

all in debt as a share of in ome. The index equals 100 in 1932Q2 to 1933Q1, the year leading

up to Ameri a's departure from the gold standard. After de lining for a de ade, nominal

debt began to rise in 1931, while real debt started to in rease a year earlier, due to de�ation.

From 1933Q2 until 1940Q2, the par value of nominal debt rose 112 per ent, while real debt

rose 82 per ent. The ratio of these indexes rea hed its nadir when the ountry left gold and

then rose 19 per ent by 1940Q2, but 22 per ent just before the 1937�1938 re ession. Those

hanges in the ratio measure how mu h debt was devalued by a higher pri e level.

19

More striking is the right panel of the �gure. The debt-GNP ratio, whether measured at

par or market value of debt, rose sharply from 15 per ent in 1930 to 42 per ent at the time

gold was abandoned. Then it hovered around 40 per ent for the next six years, until the

re ession raised the ratio. In the last few years of the de ade, when Roosevelt abandoned

the unba ked �s al expansion poli y, the debt-GNP ratio rose.

19

These numbers are nearly identi al when measured in terms of the market value of debt.

19

Ja obson, Leeper, & Preston: 1933

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940

80

100

120

140

160

180

200

0.8

0.9

1

1.1

1.2

1.3

Indexes of Gross Debt

Nominal

Real

Real/Nominal (right)

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 194010

15

20

25

30

35

40

45

50Gross Debt as Percent of GNP

Par Value

Market Value

Figure 8: Par value of U.S. gross debt, real debt is par value de�ated by GNP de�ator.

Converted to index numbers 100=1932Q2�1933Q1 (year before departure from gold stan-

dard). Nominal/Real is ratio of the two index numbers onverted to per ent. Par and market

values of debt as per entage of nominal GNP. Verti al line marks when the United States

abandoned the gold standard. Sour es: Authors' al ulations, Balke and Gordon (1986).

Figure 9 performs the a ounting exer ise that breaks the growth rate of the debt-GNP

ratio in �gure 8, Bt/PtYt, into growth rates of the three omponents. All three drove debt-

output in the three years before Roosevelt took o� e. From the �rst quarter of 1993 on,

nominal debt ontributed to driving the ratio higher. That in�uen e, though, was o�set by

higher pri es and real GNP, with the ex eption of the re ession of 1937�38.

5.5 Returns on Treasury Bond Portfolio

To interpret data related to the government's bond portfolio, we require some notation.

20

With a omplete and general maturity stru ture, the government's budget identity is

∞∑

j=0

(QD

t (t+ j) + IPt(t+ j))Bt−1(t + j) = Ptst +

∞∑

j=1

QDt (t+ j)Bt(t+ j) (4)

where QDt (t) ≡ 1 and IPt(t+j) is the interest payable on bonds outstanding at t that mature

in t+j. QDt (t+j) is the dirty pri e of bonds, de�ned as the lean pri e plus a rued interest.

The market value of debt outstanding in period t is

PMt BM

t ≡

∞∑

j=1

QDt (t+ j)Bt(t+ j) (5)

so the budget identity may be rewritten as

RMt PM

t−1BMt−1 = Ptst + PM

t BMt (6)

20

Appendix A.3 details the de�nitions and al ulations that follow.

20

Ja obson, Leeper, & Preston: 1933

Decomposition of Growth Rate of Debt-to-GNP

Rising debt, falling price level and output

Falling debt, rising price level and output

1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940-30

-20

-10

0

10

20

30

40

50

1/Y

1/P

B

B/(PY)

Figure 9: The four-quarter per entage hange in debt-GNP ratio (solid line) de omposed

into per entage hanges of its omponents: nominal debt, the inverse of the pri e level, and

the inverse of real GNP. Sour es: Balke and Gordon (1986), Hall and Sargent (2015), and

authors' al ulations.

or, in real terms

rMt PMt−1b

Mt−1 = st + PM

t bMt (7)

where bMt ≡ BMt /Pt is the real par value of debt outstanding at t. The nominal and real

rates of return on the portfolio�RMt and rMt �re�e t ex-post returns.

With BMt the par value of debt and PM

t BMt the market value, PC

t BMt−1 is the arry-over

market value of debt. The growth rate in the market value of debt may be written as

PMt BM

t

PMt−1B

Mt−1

≡PCt BM

t−1

PMt−1B

Mt−1

︸ ︷︷ ︸

nominal

rate of return

·PMt BM

t

PCt BM

t−1︸ ︷︷ ︸

size ratio

(8)

where PCt , de�ned in the appendix, re�e ts intermediate oupon payments and is the arry-

over pri e of the portfolio. The �rst ratio on the right side of (8) is the nominal return,

RMt , in (6). An ex-post real return simply de�ates the nominal return by the in�ation rate

between t− 1 and t to give rMt in (7).

The surprise omponent in the real return on the bonds portfolio is

ηt ≡ rMt −Et−1rMt (9)

This innovation an be de omposed into surprise apital gains and losses on the bond port-

folio due to in�ation and bond pri es as

ηt = RMt (1/πt − 1)︸ ︷︷ ︸

due to pri e level

+RMt

(∑∞

j=1

(Qt(t + j)−Qt−1(t+ j)

)Bt−1(t + j)

PCt BM

t−1

)

︸ ︷︷ ︸

due to bond pri es

(10)

21

Ja obson, Leeper, & Preston: 1933

Gold Standard Unba ked Fis al Expansion

Monthly Annual Monthly Annual

Nominal 0.24 2.91 0.23 2.72

Real 0.66 7.86 0.10 1.20

Surprise Real 0.40 4.81 −0.06 −0.76

Table 3: Summary of returns on government bond portfolio at monthly and annual rates.

Be ause ηt is the surprise revaluation on bonds arried into period t, its dollar magni-

tude is given by ηtPMt−1B

Mt−1. We gage the quantitative importan e of these revaluations by

omputing them as a per entage of the market value of debt at the end of period t, PMt BM

t .

Revaluation e�e ts on nominal debt are a distin t feature of an unba ked �s al expansion.

An unanti ipated in rease in the primary de� it, �nan ed by new bond issuan e, does not

trigger the expe tation of higher surpluses in the future. The new bonds raise household

nominal wealth and spending. Higher spending raises both the pri e level and produ tion;

the degree of nominal sti kiness in the e onomy determines the pre ise split between the two.

The maturity stru ture of government debt, together with how monetary poli y rea ts to

the higher in�ation, play a entral role in the resulting in�ation dynami s [Co hrane (2001),

Leeper and Walker (2013), Sims (2013), Leeper and Leith (2017)℄.

Several patterns emerge from returns data in table 3. First, nominal returns are ompa-

rable a ross the gold standard and unba ked �s al expansion period.

21

Se ond, real returns

are substantially higher in the gold standard period than in the later period (average annual

real returns of 7.86 per ent versus 1.20 per ent). Finally, on average, surprises in real returns

are strongly positive in the early period (4.81 per ent), but negative during the unba ked �s-

al expansions (−0.76 per ent).

22

These patterns are fully onsistent with surprise in�ation

devaluing government debt during Roosevelt's administration.

A key feature of an unba ked �s al expansion is that exogenous de lines in surpluses,

�nan ed by nominal debt issuan e, lead to revaluation of government debt through surprise

in reases in in�ation and de lines in bond pri es. Sims (2013) omputes surprise apital

gains and losses on U.S. government bonds sin e World War II to argue that these revalua-

tion e�e ts are important�the same order of magnitude as annual �u tuations in primary

surpluses. And Sims (2013), Leeper and Zhou (2013), and Leeper and Leith (2017) show

that surprise revaluations of debt are a generi feature of any equilibrium produ ed by jointly

optimal monetary and �s al poli ies in the presen e of distorting taxes and long-term debt.

23

Figure 10 plots the nominal and real rates of return on the government's bond portfolio

(top panel) and the one-month-ahead surprise hange in the real return. Not surprisingly,

ex-post real returns were high during the de�ation in the years before leaving gold and far

21

Return data start in 1926, so �gold standard� refers to 1926Q1 to 1933Q1.

22

Romer (1992, p. 778) estimates the ex-ante real ommer ial paper rate to �nd that it is negative nearly

the entire unba ked �s al expansion period ex ept the 1937�1938 re ession.

23

Of ourse, any sto hasti model with monetary and �s al poli y in whi h in�ation and interest rates

�u tuate will generate revaluation e�e ts. This holds regardless of the monetary-�s al poli y regime, so

merely �nding revaluation e�e ts during the re overy of the 1930s does not imply that the United States

experien ed an unba ked �s al expansion. Su h an inferen e requires identifying assumptions, whi h we turn

to in se tion 6.

22

Ja obson, Leeper, & Preston: 1933

lower on e in�ation pi ked up. But the bottom panel shows that surprise devaluations of the

bond portfolio�ηt de�ned in (9)�were a distin t feature of the unba ked �s al expansion

period.

24

1926 1928 1930 1932 1934 1936 1938 1940

-1

0

1

2Nominal and Real Returns on Bond Portfolio

Nominal

Real

1926 1928 1930 1932 1934 1936 1938 1940

-1

0

1

Innovations in Real Returns on Bond Portfolio

Figure 10: Quarterly averages of nominal and real net monthly returns on federal government

bond portfolio and one-step-ahead unanti ipated real monthly returns. See appendix A.3 for

details. Verti al line marks when the United States abandoned the gold standard. Sour e:

Hall and Sargent (2015), CRSP, and authors' al ulations.

Surprise real returns on government debt are quantitatively important. Figure 11 shows

that as a per entage of the market value of outstanding debt, these revaluations are a entral

feature of �s al �nan ing. The �gure also makes lear that after leaving the gold standard,

these revaluations are both large and frequently negative.

The de omposition of surprise real returns, graphed in �gure 12, on�rms that before

leaving the gold standard, high realized real returns were driven by low in�ation. The

negative spike due to bond pri es in 1931Q4 was reated by the Fed's e�orts to defend the

gold parity by sharply raising dis ount rates. In the period of unba ked �s al expansions,

again with the ex eption of the jump in early 1938, surprise devaluations of debt due to

in�ation dominate the surprise real returns.

The last informal pie e of empiri al eviden e about the unba ked �s al expansion ap-

pears in �gure 13, whi h plots the relative pri e of the bond portfolio. This relative pri e is

omputed as the real market value of debt over the par value of debt, whi h yields PMt /Pt,

the goods-pri e of government bonds. Bonds be ame in reasingly ostly in terms of goods

throughout the gold standard period, rea hing a peak in 1933Q1. With the departure from

24

Inspe tion of �gure 10 may suggest that ηt = rMt − 1 indi ating that innovations in real returns on

the bond portfolio are a linear transformation of real returns. Appendix A.3 shows that when taking into

onsideration oupon payments and a rued interest, η 6= rMt − 1.

23

Ja obson, Leeper, & Preston: 1933

1926 1928 1930 1932 1934 1936 1938 1940

-1.5

-1

-0.5

0

0.5

1

1.5

Surprise Revaluation as Fraction of Debt

Figure 11: Surprises in real returns on bond portfolio as per entage of market value of

outstanding debt, omputed as ηtPMt−1B

Mt−1/P

Mt BM

t . See appendix A.3 for details. Verti al

line marks when the United States abandoned the gold standard. Sour e: Hall and Sargent

(2015), CRSP, and authors' al ulations.

1926 1928 1930 1932 1934 1936 1938 1940

-1

-0.5

0

0.5

1

Decomposition of Innovation in Real Returns

Due to Inflation

Due to Bond Prices

Figure 12: De omposition of surprises in real returns on bond portfolio into omponents

due to unanti ipated in�ation and unanti ipated bond pri es. See appendix A.3 for details.

Verti al line marks when the United States abandoned the gold standard. Sour e: Hall and

Sargent (2015), CRSP, and authors' al ulations.

24

Ja obson, Leeper, & Preston: 1933

1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4Relative Price of Bond Portfolio

Figure 13: Relative pri e of the bond portfolio is the ratio of the real market value of debt to

the par value of debt, roughly equivalent to the real �pri e� of the bond portfolio. Verti al line

marks when the United States abandoned the gold standard. Sour e: authors' al ulations.

gold ame a steady devaluation of the bond portfolio, bottoming out in the middle of 1937

when the 1937�1938 re ession began. This heapening of bonds is onsistent with bondhold-

ers substituting out of debt and into buying goods and servi es�an in rease in aggregate

demand triggered by unba ked �s al expansion.

6 Stru tural VAR Analysis

We turn now to more formal analysis of �s al and monetary impa ts over the period of

unba ked �s al expansions. Be ause the identi�ed VAR methodology is well understood, we

review it only brie�y here.

25

6.1 VAR Methods

If yt is a k × 1 ve tor of time series, the e onomi stru ture is

A0yt = A+(L)yt−1 + εt (11)

where Eεtε′

t = I and εt is un orrelated with ys for s < t. The εt's are e onomi ally inter-

pretable exogenous disturban es. The redu ed-form is

yt = B(L)yt−1 + ut (12)

where, assuming that A0 is invertible, B(L) = A−10 A+(L), ut = A−1

0 εt, and Eutu′

t =A−1

0 (A−10 )′ = Σ.

25

See Leeper, Sims, and Zha (1996) or Christiano, Ei henbaum, and Evans (1999) for detailed surveys.

25

Ja obson, Leeper, & Preston: 1933

6.2 Data and Identifi ation

We estimate a seven-variable monthly VAR from April 1933 to June 1940. The seven vari-

ables are: the ommer ial paper rate, R, (NSA), the monetary base, M , (NSA), federal

primary surplus, S, (SA), the market value of nominal gross federal government debt, B,(NSA), the monetary gold sto k, G, (NSA), monthly interpolated GNP de�ator, P , (100 =1926), and monthly interpolated real GNP, Y .26

VAR estimates employ the Sims and Zha (1998) prior, whi h allows for unit roots and

ointegration, and probability bands are omputed as in Sims and Zha (1999). When restri -

tions are imposed on lagged variables, estimation follows Cushman and Zha (1997) and Zha

(1999). All variables ex ept the primary surplus and the interest rate are logged; the interest

rate is divided by 100 to put it in per entage units. We in lude six lags and a onstant.

27

This identi� ation aims to be onsistent with a tual poli y behavior in the post-gold

standard period of the 1930s. In what follows, restri tions are imposed only on A0, the on-

temporaneous intera tions among innovations in variables, leaving lags unrestri ted. With

monthly time series, this means every variable responds to past values of every other variable.

Money Supply : The supply of monetary base, Ms, depends on the short-term nominal

interest rate, R, and the monetary gold sto k, G. The de ision about whether or not to

sterilize gold in�ows lay with the Treasury during this period, but in the ase when in�ows

were not sterilized, there was a dire t impa t of G on Ms. In addition, the Federal Reserve

might de ide to adjust supply in order to in�uen e interest rates, so we have the relation

a1Mst = a2Rt + a3Gt + εMP

t (13)

Money Demand : The demand for base money in a derived demand. Demand for for real

balan es, Md − P , depends on the short-term nominal interest rate and in ome, Y

a4Mdt = a4Pt + a5Rt + a6Yt + εMD

t (14)

The identi� ation restri ts the oe� ients on nominal money and the pri e level to be equal.

Fis al Poli y : Fis al poli y hooses the primary surplus, S. An unindexed tax ode

makes revenues depend on the pri e level. Be ause surplus movements in the period were

dominated by FDR's �emergen y spending� programs, whi h were a rea tion to prevailing

e onomi onditions, there was little ontemporaneous rea tion of �s al hoi es to variables

other than measures of the pri e level and real e onomi a tivity. This leads to the �s al

rule

a7St = a8Pt + a9Yt + εPSt (15)

26

Primary surpluses were seasonally adjusted using the X-11 pro edure in RATS. The de�ator and real

GNP were interpolated from Balke and Gordon's (1986) quarterly series using the Chow and Lin (1971) al-

gorithm. Monthly series used to interpolate the de�ator in luded M2, the onsumer pri e index, the whole-

sale pri e index, the long-term yield on Treasury bonds (NBER Ma rohistory Database, m13033a), and

index omposite wages (NBER Ma rohistory Database, m08061 ); series used to interpolate real GNP in-

luded industrial produ tion, omposite index of six roughly oin ident series (NBERMa rohistory Database,

m16003a); index of fa tory employment, total durable goods (NBER Ma rohistory Database, m08146a), and

produ tion worker employment, manufa turing (NBER Ma rohistory Database, m08010b).

27

In notation analogous to that in Sims and Zha (1998), these results set the hyperparameters for the prior

as µ1 = 0.6, µ2 = 0.3, µ3 = 1.0, µ4 = 1.75, µ5 = 2.0, µ6 = 2.0. The prior was hosen based on the model's

marginal data density.

26

Ja obson, Leeper, & Preston: 1933

Government Debt : The VAR in ludes the nominal market value of gross federal debt,

B. In prin iple, bond pri es rea t immediately to all sho ks in the e onomy, so B is an

�information variable,� in Leeper, Sims, and Zha's (1996) terminology. The debt equation is

a10Bt = a11Rt + a12Mt + a13St + a14Gt + a15Pt + a16Yt + εBt (16)

Gold : With the passage of the Gold Reserve A t in January 1934, the Treasury bought

all gold at the pri e hosen by the Treasury and the President, whi h was $34.00 an oun e,

a devaluation of the gold-value of the dollar of almost 60 per ent from its value over the

previous entury. This made the demand for gold perfe tly elasti at that pri e. Supply of

gold to Ameri a, on the other hand, was driven by both exogenous politi al onditions in

Europe and endogenous fa tors within the United States. Among those endogenous fa tors

were the relative strength of the U.S. re overy, U.S. willingness to buy unlimited quantities

of gold at a high pri e, in reased sale of U.S. mer handise abroad as the dollar depre iated,

the in�ow of apital to the United States, and foreign-owned apital sent to U.S. to build

up dollar balan es or to pur hase Ameri an se urities [Paris (1938)℄. We model the supply

of monetary gold as a fun tion of the nominal interest rate and goods-market onditions.

Rather than separating demand and supply of gold, we posit an expression for the equilibrium

monetary gold sto k

28

a17Gt = a18Rt + a19Pt + a20Yt + εGt (17)

Goods Market : The remaining variables in the VAR are the pri e level and real GNP,

whi h we refer to as �goods market variables.� We follow mu h of the VAR literature by

treating these as inertial variables that are predetermined and obey a re ursive ordering. The

limitation in this assumption is that we do not distinguish between the two �goods market

sho ks,� treating them simply as disturban es unrelated to the behavior identi�ed in other

equations

a21Pt = a22Yt + εPt (18)

a23Yt = εYt (19)

Predeterminedness of goods market variables is not a stringent restri tion for data at a

monthly frequen y.

Table 4 summarizes the identi� ation. With 28 distin t moments in the ovarian e matrix

of innovations and 23 freely estimated parameters, the system is overidenti�ed.

The identi� ation determines the money sto k, nominal interest rate, and the gold sto k

simultaneously. With P and Y predetermined, given (18) and (19), (15) implies S. Equilib-rium in the money and gold markets jointly determines M , R, and G. Finally, the market

value of debt emerges from (16).

Table 5 reports posterior modes and 68-per ent probability intervals for the estimated

parameters in table 4's pattern matrix. The money supply rule is onsistent with the entral

bank expanding high-powered money in response to surprise in reases in the nominal interest

rate. Contemporaneous intera tions between gold and the base are weak. Money demand has

a signi� antly negative interest elasti ity and essentially no short-run in ome elasti ity, whi h

28

Equilibrium emerges from equating gold demand, Gd = f(PG), and gold supply, Gs = g(PG, R, P, Y ),where PG

is the pegged gold pri e, and solving out for PG.

27

Ja obson, Leeper, & Preston: 1933

MP MD FP B G P Y

R × × × ×

M × ×1 ×

S × ×

B ×

G × × ×

P ×1 × × × ×

Y × × × × × ×

Table 4: Pattern matrix for Baseline identi� ation. × denotes a freely estimated parameter,

×1 are restri ted to be of equal but opposite sign, and a blank is a zero restri tion.

is not surprising in monthly data. Primary surpluses are weakly onne ted to goods market

innovations, although over longer horizons real e onomi a tivity does a�e t surpluses. Real

in ome innovations raise the monetary gold sto k, whi h is onsistent with the U.S. e onomi

re overy indu ing gold in�ows from abroad, whi h are met by an elasti demand for gold

by the Treasury. Finally, the nominal market value of government bonds is signi� antly

asso iated with ontemporaneous innovations in variables, re�e ting the responsiveness of

asset pri es to news. Those ontemporaneous relationships make good e onomi s sense: a

surprisingly high market value of bonds is asso iated with negative innovations in the interest

rate, money sto k, primary surpluses, and the pri e level, but positive innovations in gold.

Strongest e�e ts are asso iated with the interest rate and in�ation, as theory would suggest.

.062Ms

(.021,.070)= 1.134R

(.597,1.836)+ .001G

(−.005,.002)+ εMP

.0586(Md − P )(.0349,.0986)

= −1.844R(−2.042,−.746)

+ .004Y(−.004,.011)

+ εMD

.0046S(.0042,.0049)

= −.009P(−.030,.009)

+ .003Y(−.011,.004)

+ εPS

.019G(.016,.020)

= .135R(−.270,.516)

+ .010P(−.010,.029)

+ .010Y(.002,.018)

+ εG

.090B(.083,.097)

= −.782R(−1.020,−.525)

− .024M(−.033,−.016)

− .0021S(−.0026,−.0015)

+ .006G(.004,.008)

− .034P(−.054,−.015)

+ .006Y(−.002,.013)

+ εB

.167P(.155,.180)

= .038Y(.019,.055)

+ εP

.066Y(.061,.072)

= εY

Table 5: Posterior mode estimates of parameters in table 4's pattern matrix. 68-per ent

probability intervals appear in parentheses. Coe� ients and probability intervals in the

table are divided by 1000.

28

Ja obson, Leeper, & Preston: 1933

6.3 Primary Surplus Impa ts

Figure 14 reports the dynami impa ts of a surprise de rease in the real primary surplus

during the unba ked �s al expansion period. The one standard deviation initial sho k raises

the primary de� it by $0.22 billion, whi h is about half of the average annualized monthly

de� it in the sample. Be ause the de� it de ays rapidly, the total in rease over the three-

year fore ast horizon is only $0.52 billion. This is a relatively small and transitory �s al

impulse. Higher de� its do not bring forth higher future surpluses, lending support to the

interpretation that �s al expansion is unba ked.

Higher de� its produ e Keynesian impa ts. Pri es and output, whi h the identi� ation

prevents from rising ontemporaneously, steadily in rease and signi� antly so. The monthly

impa ts peak at 0.0046 per ent for the pri e level and 0.0098 for real GNP, but the persisten e

of the responses implies that the total in reases over the three-year horizon are substantial:

0.12 per ent for the pri e level and 0.26 per ent for output.

Monetary poli y makes e�ort to o�set the in�ationary onsequen es of the �s al expan-

sion, suggesting the Fed behaves passively. Nominal interest rates fall slightly in the short

run. The lower nominal rates, together with higher expe ted in�ation, drive ex-ante real

rates lower. Lower real rates indu e households and �rms to shift demand for goods into the

present.

New nominal bonds �nan e the higher de� its. Debt jumps on impa t and remains

elevated. E onomi re overy en ourages gold to �ow into the United States. By hoosing not

to sterilize gold in�ows, the Treasury allows the monetary base to expand to a ommodate

rising demand for money.

Looking down the olumn in �gure 14 it is easy to see the onventional monetary narra-

tive of the re overy that Friedman and S hwartz (1963), Romer (1992), and Steindl (2004)

re ount.

29

The initial revaluation of gold, together with the steady in�ows of gold largely

due to politi al un ertainty in Europe, were permitted by the Treasury to steadily in rease

the monetary base. Expansion in high-powered money stimulated real a tivity and raised

pri es. At the same time, enhan ed on�den e in banks after the early 1930s rises redu ed

ash hoarding and raised the in ome velo ity of money to reinfor e the expansionary e�e ts

of the growth in the base.

But the impulse responses reate a problem for this onventional narrative. How does

one re on ile monetary-indu ed e onomi re overy with the sharp short-run de lines in pri-

mary surpluses and the persistent in rease in nominal government debt? Existing literature

does not address this question, primarily be ause the �s al dimensions have not been fully

integrated with the monetary interpretations of the re overy.

29

Friedman and S hwartz give this narrative a di�erent twist than Romer. Friedman and S hwartz (1963,

p. 499) write that �. . . the rise in the money sto k [from 1933 to 1937℄ was produ ed not by the monetary

authorities but by gold in�ow. Though a idental gold in�ows served the same e onomi fun tion as om-

pliant monetary authorities would have, it o urred despite rather than be ause of the a tions of unions,

business organizations, and government in pushing up pri es.� Romer, in ontrast, attributes mu h of the

growth in base money to the Treasury's de ision not to sterilize the in�ows, whi h was a poli y hoi e.

29

Ja obson, Leeper, & Preston: 1933

5 10 15 20 25 30 35

-3-2-10

×10-4 Interest Rate

5 10 15 20 25 30 350

0.01

0.02

Base

5 10 15 20 25 30 35

-0.2

-0.1

0Surplus

5 10 15 20 25 30 350

0.005

0.01

Debt

5 10 15 20 25 30 350

0.02

0.04

Gold

5 10 15 20 25 30 350

2

4

6

×10-3 Deflator

5 10 15 20 25 30 350

0.01

0.02GNP

Figure 14: Responses to an unanti ipated de rease in the primary surplus in the unba ked

�s al expansion period (April 1933 to June 1940). Solid lines are maximum likelihood es-

timates; dashed lines are 68 per entile probability bands based on 1000 draws from the

posterior distribution of all the VAR parameters.

30

Ja obson, Leeper, & Preston: 1933

6.4 Monetary Impa ts

Higher de� its generate positive omovements among output, the pri e level, the monetary

base, and the gold sto k. But that interpretation as ribes to �s al poli y a ausal role.

Perhaps those �s ally-indu ed orrelations are but a small part of the story about the re ov-

ery. Perhaps other disturban es, unrelated to �s al poli y, generate the same omovements,

but a ount for the bulk of �u tuations in output and pri es, as the onventional monetary

narrative maintains.

We address these on erns by examining the remaining impulse response fun tions. Fig-

ure 15 reports the dynami impa ts of four sho ks related to the monetary se tor�monetary

poli y, money demand, �government debt,� and �the gold sto k.� Our identi� ation does not

atta h any distin t behavioral interpretation to the sho ks in the equations for debt and

gold, other than that the disturban es emanate from bond and gold markets.

From early 1933 until De ember 1936, the Treasury opted not to sterilize gold in�ows,

whi h permitted the monetary base to expand along with the gold sto k. We view �gure 15

with an eye toward sho ks that move base money strongly and persistently. The �rst two

olumns�monetary poli y and money demand disturban es�generate su h movements, but

only money demand raises the gold sto k, and then does so only very brie�y. In any ase,

neither sho k has signi� ant impa ts on the pri e level.

Turning to the �fth row of the �gure�responses of the gold sto k�we see that both

bond-market and gold-market sho ks persistently raise the gold sto k, with gold-market

sho ks quantitatively more important. Positive innovations in gold are followed by a higher

monetary base, although not signi� antly higher; if anything, though, higher monetary gold

leads to lower pri es and real GNP. These disturban es tend to be followed by a lower

ommer ial paper rate and a higher market value of government debt.

Only disturban es to the primary surplus generate the full set of movements in assets,

the pri e level, and real GNP that would seem to align with existing explanations of the

re overy. But in the VAR, those movements are initiated by an exogenous shift in �s al

behavior. These impa ts of a sho k that raise the primary de� it are fully onsistent with

what the theory predi ts for the onsequen es of an unba ked �s al expansion. We turn now

to how important these �s al disturban es are in generating �u tuations in the variables of

interest.

6.5 Quantitative Importan e

We examine varian e and histori al de ompositions to assess the quantitative importan e of

�s al poli y for the e onomi re overy. Those de ompositions measure how important ea h

exogenous sho k is for future movements in the variables in the VAR.

6.5.1 Varian e De ompositions Table 6 reports varian e de ompositions of the seven

variables in the VAR at 6- and 36-month horizons. These statisti s re ord how important

disturban es in ea h exogenous sho k are for explaining �u tuations in the variables, on

average over the estimated sample.

Looking �rst at the goods market variables, P and Y , in the �rst two panels, aside from

own sho ks, the only disturban e that a ounts for an important fra tion of error varian e in

31

Ja obson,Leeper,&Preston:1933

10 20 300

2

4×10-4

MP

Interest Rate

10 20 30

-0.015

-0.01

-0.005

Base

10 20 30

-0.01

0

0.01

0.02Surplus

10 20 30

-2024

×10-3 Debt

10 20 30

-0.01

0

0.01

Gold

10 20 30

-3-2-101

×10-3 Deflator

10 20 30

-5

0

5×10-3 GNP

10 20 30

-4

-2

0×10-4

MD

Interest Rate

10 20 30

-15

-10

-5

0×10-3 Base

10 20 30

-0.010

0.010.020.03

Surplus

10 20 30

-20246

×10-3 Debt

10 20 30

-0.01

0

0.01

Gold

10 20 30-3-2-101

×10-3 Deflator

10 20 30

-5

0

5×10-3 GNP

10 20 30-15

-10

-5

0

×10-5B

Interest Rate

10 20 30

-4

-2

0

2

×10-3 Base

10 20 30

0

10

20

×10-3 Surplus

10 20 300

0.005

0.01

Debt

10 20 30-5

0

5

10

×10-3 Gold

10 20 30

-1

0

1×10-3 Deflator

10 20 30-6

-4

-2

0

×10-3 GNP

10 20 30

-2-101

×10-4G

Interest Rate

10 20 30

0

5

10

×10-3 Base

10 20 30

-0.01

0

0.01

0.02Surplus

10 20 30-5

0

5×10-3 Debt

10 20 300

0.05

Gold

10 20 30

-4

-2

0×10-3 Deflator

10 20 30

-10

-5

0

5×10-3 GNP

Figure 15: Responses to a unanti ipated sho ks in the �monetary se tor,� whi h in ludes monetary poli y (MP), money demand

(MD), government debt (B), and the gold sto k (G). Unba ked �s al expansion period (April 1933 to June 1940). Solid lines

are maximum likelihood estimates; dashed lines are 68 per entile probability bands based on 1000 draws from the posterior

distribution of all the VAR parameters.

32

Ja obson, Leeper, & Preston: 1933

those variables omes from �s al poli y. A bit under 20 per ent of goods market variables'

�u tuations arise from sho ks to the primary surplus. Monetary disturban es�monetary

poli y, money demand, and gold �ows�jointly explain only 5.6 per ent.

Money market sho ks together a ount for substantial fra tions of error varian es in

the monetary base (53 per ent) and the ommer ial paper rate (81 per ent). But primary

surpluses explain almost all the remaining varian e in base money (42 per ent), suggesting

a strong endogenous response of money to �s al disturban es.

Primary surpluses�the �fth panel�are largely exogenous, with own sho ks a ounting

for 98 per ent of surplus movements at all horizons. This �nding is onsistent with Roo-

sevelt's �emergen y spending� driving �s al poli y in the period. Of ourse, this spending

was most de idedly not exogenous in the usual meaning of the term be ause the spending

was an expli it response to e onomi onditions in the pre eding years.

Primary surplus disturban es explain one-third of the fore ast error varian e in gold. This

�nding belies the argument by Friedman and S hwartz (1963) and others that gold in�ows

were almost entirely due to European politi al turmoil and gold dis overies. Of ourse,

a substantial fra tion (60 per ent) of �u tuations in gold are due to exogenous sho ks in

demand and supply for gold, whi h may re�e t the fa tors that Friedman and S hwartz

emphasize.

6.5.2 Histori al De ompositions The ve tor of variables in the VAR, yt, may be

de omposed into the fore ast onditional only on initial onditions using estimated VAR

parameters, E0yt, and the sum of the realized exogenous sho ks, εt, as

yt =

t−1∑

s=0

Csεt−s + E0yt (20)

Group the sho ks into three bins: �s al poli y, εFt = εPSt , goods markets, εMt = (εPt , ε

Yt ),

and other, εOt = (εMPt , εMD

t , εGt , εBt ), with asso iated moving-average oe� ients CF

, CM,

and CO. Then (20) for variable j in period t may be written as

yjt = E0yjt +

t∑

i=1

CFj (i)ε

Fi +

t∑

i=1

Cj(i)εMt +

t∑

i=1

Cj(i)εOt (21)

where ea h summation is the umulative impa t of exogenous sho ks on variable j from

period 1 to period t.Figures 16 and 17 plot all the omponents in de omposition (21) for the pri e level and

real GNP. After a ounting for lags in the VAR estimation, fore asts run from O tober 1933

through June 1940. Solid lines are a tual values, yjt, and solid dotted lines are fore asts,

E0yjt. The remaining three lines are a tual values less the ontributions of ea h of the three

sho k groups.

Fore asts of both variables rise monotoni ally over the period, suggesting that in the

absen e of sho ks, deterministi dynami s would raise pri es and output. The marginal

ontribution of ea h sho k group appears as the verti al distan e between the a tual value

and the value less that group's addition. A onsistent pattern a ross both �gures is that

the four sho ks that onstitute the �other� group�monetary poli y, money demand, gold,

33

Ja obson, Leeper, & Preston: 1933

Per ent of P Due to Sho ks in

Months MP MD FP B G P Y

6 0.5 0.2 9.2 0.1 0.4 89.1 0.6

36 1.1 0.3 18.1 0.0 4.2 75.6 0.7

Per ent of Y Due to Sho ks in

Months MP MD FP B G P Y

6 0.0 1.5 8.9 0.3 0.0 6.8 82.5

36 0.1 1.2 17.4 0.7 0.7 4.6 75.3

Per ent of M Due to Sho ks in

Months MP MD FP B G P Y

6 60.2 13.1 20.8 0.0 1.4 1.4 3.0

36 46.2 6.7 41.7 0.0 2.1 0.7 2.6

Per ent of R Due to Sho ks in

Months MP MD FP B G P Y

6 26.5 57.6 4.2 1.4 3.2 4.5 2.7

36 27.1 53.9 7.4 1.6 1.7 6.2 2.3

Per ent of PS Due to Sho ks in

Months MP MD FP B G P Y

6 0.1 0.7 98.0 0.4 0.1 0.3 0.4

36 0.1 0.7 97.9 0.4 0.1 0.3 0.4

Per ent of G Due to Sho ks in

Months MP MD FP B G P Y

6 0.2 0.6 19.8 1.0 72.2 3.2 3.1

36 0.1 0.5 33.6 0.7 58.8 4.6 1.7

Per ent of B Due to Sho ks in

Months MP MD FP B G P Y

6 0.2 5.4 22.3 61.5 5.4 1.5 3.7

36 0.4 2.6 26.9 63.8 1.9 0.9 3.6

Table 6: Per entage of fore ast error varian e in GNP de�ator (P ), real GNP (Y ), monetary

base (M), ommer ial paper rate (R), monetary gold sto k (G), and nominal market value

of debt (B) attributable to sho ks to ea h equation. Columns may not sum to 100 due to

rounding.

34

Ja obson, Leeper, & Preston: 1933

Figure 16: Histori al de omposition of the pri e level into the right-hand-side omponents

of equation (21).

Figure 17: Histori al de omposition of real GNP into the right-hand-side omponents of

equation (21).

and debt�have small e�e ts that run ounter to Roosevelt's e onomi obje tives: pri es and

output would be a bit higher in the absen e of those disturban es.

Fis al sho ks always serve to raise real GNP and tend to raise the pri e level, ex ept for

a period in 1938�1939. Goods market disturban es are the biggest ontributors to ma roe-

onomi a tivity, but their impa ts an be positive or negative, depending on the period.

35

Ja obson, Leeper, & Preston: 1933

Figure 18: Per entages of fore ast error of the pri e level and real GNP due to sho k at-

egories in equation (21), O tober 1933�June 1940, using April�September 1933 as initial

onditions. Extreme values asso iated with near zero fore ast errors have been ex luded.

A learer sense of ea h sho k group's importan e may be gleaned from the per entage of

fore ast errors a ounted for by the group. Computing

1 =

∑t

i=1CFj (i)ε

Fi

yjt −E0yjt+

∑t

i=1Cj(i)εMt

yjt −E0yjt+

∑t

i=1Cj(i)εOt

yjt − E0yjt(22)

and onverting to per entages, we obtain �gure 18. The �gure ex ludes periods when per-

entages are extreme be ause fore ast errors are lose to zero.

30

Fis al poli y disturban es frequently a ount for a substantial fra tion of fore ast errors

in the pri e level, rising steadily in the early part of the sample to explain about 100 per ent

in early 1936 (top panel of �gure). That per entage rises still further later in 1936. On

average, between January 1934 and July 1938, surplus disturban es a ount for over 45

per ent of pri e-level fore ast errors. In some periods, surpluses play a bigger role than do

sho ks from goods markets. In ontrast, through the middle of 1938, �other� sho ks always

drive the pri e level down.

The story of �s al sho ks for output is more varied. Until late 1937, �s al poli y on-

sistently drove output above fore ast, a ounting for an average of 66 per ent of real GNP

fore ast errors between January 1934 to November 1937. By the se ond half of the re ession

that started in May 1937, though, �s al sho ks were driving down output substantially.

30

Those periods arise when the a tual values are lose to the fore asts. Removing periods in whi h at

least one per entage ex eeds 200, for the pri e level removes Mar h�July 1936 and O tober�De ember 1938,

while for real GNP removes January 1934, O tober�De ember 1934, De ember 1937, July 1938, Mar h�April

1939, July�August 1939, and Mar h�April 1940.

36

Ja obson, Leeper, & Preston: 1933

7 Lessons for Today

We have argued that unba ked �s al expansion was the sour e of the re overy from the Great

Depression. Roosevelt's �try anything� poli ies produ ed debt-�nan ed primary de� its that

remained in pla e until re overy was underway. Monetary poli y ombined with that �s al

poli y to stabilize debt by preventing nominal interest rates from rising with in�ation. The

paper o�ered a variety of eviden e that debt-�nan ed de� its generated gold in�ows and ex-

panded the monetary base at the same time that they raised pri es and output. Gold in�ows

and higher base money that were not asso iated with higher de� its and nominal debt have

little predi tive power for the GNP de�ator and real GNP. Despite rapid growth in nominal

debt between 1933 and 1937, the debt-GNP ratio was stable at about 40 per ent, the level

it had rea hed before the United States abandoned gold. This leads to the on lusion that

unba ked �s al expansion lifted the U.S. e onomy out of the depression without endangering

the reditworthiness of the ountry.

Roosevelt's su essful, if in omplete, re�ation arries two important lessons for poli y-

makers today. Many ountries now su�er from low�below-target�in�ation rates and tepid

e onomi growth. Rather than relying on a joint monetary-�s al atta k on the problem, as

Roosevelt did, these ountries are leaning entirely on monetary poli y. Central banks in the

Euro Area, Sweden, Switzerland, and Japan have set poli y interest rates below zero and

undertaken large-s ale asset pur hases in an e�ort to redu e real interest rates and stimulate

aggregate demand and in�ation. This poli y relies on intertemporal substitution indu ed by

low real rates, rather than the wealth e�e ts of an unba ked �s al expansion. Fis al poli ies

in those areas, meanwhile, have la ked Roosevelt's initial single-minded goal to stimulate

the e onomy, �u tuating between �s al stimulus and �s al austerity. Despite the Her ulean

e�orts of monetary authorities for several years, there is little eviden e of re�ation in those

ountries.

Ironi ally, those same ountries and the United Kingdom, like the United States in the

1930s, are well positioned to undertake unba ked �s al expansions. Monetary poli ies are

already passive and entral banks are on board to a hieve higher in�ation rates.

31

A se ond lesson from the Roosevelt poli ies is that �s al stimulus and �s al sustainability

need not be in on�i t. When the aim is to raise in�ation and e onomi growth, higher nom-

inal government debt�if people are onvin ed it does not portend higher future taxes� an

a hieve both the ma roe onomi obje tives and the goal of stabilizing debt. The two goals

go hand-in-hand: higher in�ation redu es the real value of the debt and higher e onomi

growth raises surpluses and redu es debt-output ratios. But to engineer an unba ked �s-

al expansion, governments must understand that rapid growth in nominal debt need not

threaten �s al sustainability, just as it didn't in 1930s Ameri a.

In the urrent atmosphere of what Sims (2016) alls �hyper-Ri ardian� beliefs about poli y

in whi h the publi sees higher debt as bringing forth mu h higher surpluses in the future, it

may be di� ult for poli ymakers to redibly ommit to an unba ked �s al expansion. Here,

too, FDR may have something to tea h. Roosevelt never laimed to be aiming for what

even he might have regarded as �irresponsible� �s al poli y. But his ommuni ations and

31

Be ause individual Euro Area ountries do not ontrol their monetary poli y, it would require a o-

ordinated unba ked �s al expansion a ross member nations together with the ECB's pegging of interest

rates.

37

Ja obson, Leeper, & Preston: 1933

a tions made lear that he was willing to do whatever it took to bring the ountry out of

the depression. Roosevelt was also agnosti , willing to experiment, even with what at the

time seemed to be radi al poli ies. He kept the publi 's attention on the poli y obje tives,

obje tives over whi h there was nearly universal agreement, rather than on the poli y tools.

Roosevelt's eventual ba ktra king on �s al stimulus also arries a valuable message for

poli y makers today. Su essful re overy from severe e onomi downturns mandates single-

minded pursuit of e onomi re overy obje tives. Allowing an illary on erns to enter the

al ulus onfuses e onomi de ision makers and an undermine that su ess.

38

Ja obson, Leeper, & Preston: 1933

Appendi es

A Data

A.1 Net Interest

A.1.1 Interest Re eipts This se tion details our sour es and al ulation of monthly

net interest. Interest re eipts are only available on a yearly basis in the Annual Report

of the Se retary of the Treasury on the State of the Finan es. From 1928 to 1940, we

use the total of series alled �Interest, ex hange, and dividends on apital sto k� or �Total

interest, ex hange, dividends� omputed from the unrevised daily Treasury statements.

32

Disaggregated omponents of this series are available in tables based on warrants issued or

revised daily Treasury statements.

33

32

From 1928 to 1933, interest re eipts are split into general and spe ial funds ategories. We use total

interest re eipts.

33

On Page 389 of the 1928 Annual Report, daily Treasury statements (unrevised) are de�ned as �gures

ompiled �from the latest daily reports re eived by the Treasurer of the United States, from Treasury o� ers,

and publi depositaries holding Government funds. The daily Treasury statement, therefore, is a urrent

report ompiled from latest available information, and, by reason of the promptness with whi h the infor-

mation is obtained and made publi , it has ome into general use as re�e ting the �nan ial operations of the

Government overing a given period, and gives an a urate idea of the a tual ondition of the Treasury as

far as it is as ertainable from day to day. This is known as ` urrent ash basis,' a ording to daily Treasury

statements (unrevised).� Revised Treasury statements re�e t a tual transa tions during the period under

review. Page 373 of the 1929 annual report explains that re eipts and expenditures are revised �on a ount of

the distan e of some of the Treasury o� es and depositaries from the Treasury, it is obvious that the report

from all o� ers overing a parti ular day's transa tions an not be re eived and assembled in the Treasury

at one time without delaying for several days the publi ation of the Treasury statement.� Warrants issued

(re eipts) are de�ned based on Se tion 305 of the Revised Statues as, �re eipts for all moneys re eived by

the Treasurer of the United States shall be indorsed upon warrants signed by the Se retary of the Treasury,

without whi h warrants, so signed, no a knowledgment for money re eived into the Publi Treasury shall

be valid. The issuan e of warrants by the Se retary of the Treasury, as provided by law, represents the

formal overing of re eipts into the Treasury.� Warrants issued (expenditures) are de�ned by the fa t that,

�The Constitution of the United States provides that no money shall be drawn from the Treasury but in

onsequen e of appropriations made by law. Se tion 305 of the Revised Statutes requires that the Treasurer

of the United States shall disburse the moneys of the United States upon warrants drawn by the Se retary

of the Treasury. As the warrants are issued by the Se retary they are harged against the appropriate

appropriations provided by law. Some of these warrants do not represent a tual payments to laimants,

but are merely advan es of funds to be pla ed to the redit of disbursing o� ers of the Government with

the Treasurer of the United States for the payment of Government obligations. The disbursing o� er then

issues his he k on the Treasurer in payment of su h obligations. As far as the appropriation a ounts are

on erned, the warrants issued and harged thereto onstitute expenditures, but it will be observed that

su h expenditures ne essarily in lude unexpended balan es to the redit of the disbursing o� ers. Under

normal onditions these balan es over a period of several years �u tuate very little in the aggregate, and the

di�eren e between the total expenditures on a warrant basis and a ash basis (revised) is immaterial.

39

Ja obson, Leeper, & Preston: 1933

Figure 19: 1928 Annual Report, page 391

Figure 20: 1929 Annual Report, page 374

In 1927, interest re eipts are only available based on warrants issued.

34

Although the

aggregate total of �Interest, premium, and dis ount� is no longer provided, the disaggregated

elements of this total are in luded. We ontinue to in luded dividends, premiums, dis ounts,

and ex hanges to be onsistent with the years when only the aggregate series is available.

34

See footnote 33 for a des ription of warrants versus unrevised ash basis.

40

Ja obson, Leeper, & Preston: 1933

Figure 21: 1927 Annual Report, page 431

41

Ja obson, Leeper, & Preston: 1933

Starting in 1922, interest re eipts, premium, dis ounts, and ex hanges are no longer given

as separate ategories. The omponents of federal re eipts are listed alphabeti ally.

35

Figure 22: 1922 Annual Report, page 107

Interest re eipts on foreign obligations � a subset of total interest re eipts � are available

on an unrevised ash basis. This data is also available at a monthly frequen y for �s al years

1929 to 1931 and 1936 to 1940. The lo ation of these data is in luded in Table 7.

35

Net warrants issued in ludes unexpended balan es to the redit of disbursing o� ers at the end of the

year, but not expenditures under su h unexpended balan es at the beginning of the year.

42

Ja obson, Leeper, & Preston: 1933

Table name Year Basis Page number

Comparison of re eipts, �s al years 1920 and 1919 1920 warrant 262/263

Comparison of re eipts, �s al years 1921 and 1920

1921

warrant 140

Re eipts and expenditures for �s al years 1920 and 1921 (int. on foreign obligations) unrevised 152

Comparison of re eipts, �s al years 1922 and 1921

1922

warrant 107

Re eipts and expenditures for �s al years 1921 and 1922 (int. on foreign obligations) unrevised 100

Comparison of re eipts, �s al years 1923 and 1922

1923

warrant 114

Re eipts and expenditures for �s al years 1922 and 1923 (int. on foreign obligations) unrevised 107

Comparison of re eipts, �s al years 1924 and 1923

1924

warrant 131

Re eipts and expenditures for �s al years 1923 and 1924 (int. on foreign obligations) unrevised 123

Comparison of re eipts, �s al years 1925 and 1924

1925

warrant 150

Re eipts and expenditures for �s al years 1924 and 1925 (int. on foreign obligations) unrevised 141

Comparison of re eipts, �s al years 1926 and 1925

1926

warrant 429

Re eipts and expenditures for �s al years 1925 and 1926 (int. on foreign obligations) unrevised 176

Comparison of re eipts, �s al years 1927 and 1926

1927

warrant 431

Re eipts and expenditures for �s al years 1926 and 1927 (int. on foreign obligations) unrevised 30

Re eipts and expenditures for the �s al year 1928

1928

revised 391

Re eipts and expenditures for the �s al year 1928 (int. on foreign obligations) unrevised 19

Re eipts and expenditures for the �s al year 1929

1929

revised 375

Re eipts and expenditures for the �s al year 1929 (int. on foreign obligations) unrevised 20

Ordinary Re eipts (monthly) (foreign obligations) unrevised 535

Re eipts and expenditures for the �s al year 1930

1930

revised 469

Re eipts and expenditures for the �s al year 1930 (int. on foreign obligations) unrevised 35

Ordinary Re eipts (monthly) (foreign obligations) unrevised 631

Re eipts and expenditures for the �s al year 1931

1931

warrant 426

Re eipts and expenditures for the �s al year 1931 (int. on foreign obligations) unrevised 25

Re eipts and Expenditures, by months (foreign obligations) unrevised 575

Re eipts and expenditures for the �s al year 1932

1932

warrant 341

Re eipts and expenditures for the �s al year 1932 (int. on foreign obligations) unrevised 27

Details of re eipts by sour es and funds, for the �s al year 1933

1933

warrant 310

Re eipts and expenditures for the �s al year 1933 (int. on foreign obligations) unrevised 19

Details of re eipts by sour es and funds, for the �s al year 1934

1934

warrant 276

Re eipts and expenditures for the �s al year 1934 (int. on foreign obligations) unrevised 20

Details of re eipts by sour es and funds, for the �s al year 1935

1935

warrant 296

Re eipts and expenditures for the �s al year 1935 (int. on foreign obligations) unrevised 32

Details of re eipts by sour es and funds, for the �s al year 1936

1936

warrant 314

Re eipts and expenditures for the �s al year 1935 (int. on foreign obligations) unrevised 35

Classi�ed re eipts and expenditures, monthly unrevised 339/344

A tual re eipts for the �s al year 1937

1937

warrant 380

Classi�ed re eipts and expenditures for the �s al years 1932 to 1937 unrevised 338

Classi�ed re eipts and expenditures, monthly (int. on foreign obligations) unrevised 320/326

A tual re eipts for the �s al year 1937

1938

warrant 457

Classi�ed re eipts and expenditures for the �s al years 1932 to 1938 unrevised 401

Classi�ed re eipts and expenditures, monthly (int. on foreign obligations) unrevised 379/387

Details of re eipts, by sour es and a ounts

1939

warrant 314

Classi�ed re eipts and expenditures, monthly (int. foreign obligations) unrevised 337/345

Details of re eipts, by sour es and a ounts.

1940

warrant 587

Classi�ed re eipts and expenditures, monthly (int. foreign obligations) unrevised 612/619

Table 7: Table names and page numbers from the Annual Reports of the Se retary of the

Treasury for interest re eipts

43

Ja obson, Leeper, & Preston: 1933

A.1.2 Interest Expenditures Interest expenditures are available on a monthly basis

starting in January 1922. For July 1919 to De ember 1921, interest expenditures are available

on a quarterly frequen y. We divide the quarterly data by three to interpolate monthly data

for this time period.

Table name Year Basis Page number

Preliminary Statement Showing Classi�ed Expenditures (quarterly)... 1920 unrevised see 1921 357

Re eipts and expenditures of the Government for �s al (yearly)... unrevised see 1926 448

Preliminary Statement Showing Classi�ed Expenditures (quarterly)...

1921

unrevised 357

Re eipts and expenditures of the Government for �s al (yearly)... unrevised see 1926 448

Preliminary Statement Showing Classi�ed Expenditures (monthly)...

1922

unrevised 103

Re eipts and expenditures of the Government for �s al (yearly)... unrevised see 1926 448

Preliminary Statement Showing Classi�ed Expenditures (monthly)...

1923

unrevised 110

Re eipts and expenditures for �s al years 1922 and 1923 (yearly) unrevised 107

Preliminary Statement Showing Classi�ed Expenditures (monthly)...

1924

unrevised 127

Re eipts and expenditures for �s al years 1923 and 1924 (yearly) unrevised 123

Preliminary Statement Showing Classi�ed Expenditures (monthly)...

1925

unrevised 145

Re eipts and expenditures for �s al years 1924 and 1925 (yearly) unrevised 142

Expenditures of the Government, by months for the �s al year 1926

1926

unrevised 452

Re eipts and expenditures of the Government for �s al years (yearly) unrevised 450

Expenditures by months, lassi�ed a ording to...

1927

unrevised 463

Ordinary re eipts, expenditures hargeable against... (yearly) unrevised 448

Expenditures by months, lassi�ed a ording to...

1928

unrevised 425

Re eipts and expenditures for the �s al year 1928 unrevised 19

Expenditures by months, lassi�ed a ording to...

1929

unrevised 414

Re eipts and expenditures for the �s al year 1929 (yearly) unrevised 20

Expenditures by months, lassi�ed a ording to...

1930

unrevised 510

Re eipts and expenditures for the �s al year 1930 (yearly) unrevised 35

Expenditures by months, lassi�ed a ording to...

1931

unrevised 464

Ordinary re eipts, expenditures hargeable against... (yearly) unrevised 446

Expenditures by months, lassi�ed a ording to...

1932

unrevised 371

Re eipts and expenditures for the �s al year 1932 (yearly) unrevised 27

Expenditures by months, lassi�ed a ording to...

1933

unrevised 313

Re eipts and expenditures for the �s al year 1933 (yearly) unrevised 280

Expenditures by months, lassi�ed a ording to...

1934

unrevised 308

Re eipts and expenditures for the �s al year... (yearly) unrevised 305

Expenditures by months, lassi�ed a ording to...

1935

unrevised 330

Expenditures by months, lassi�ed a ording to (yearly)... unrevised 334

Classi�ed re eipts and expenditures, monthly

1936

unrevised 337

Classi�ed re eipts and expenditures, monthly (yearly) unrevised 339

Classi�ed re eipts and expenditures, monthly

1937

unrevised 322/328

Classi�ed re eipts and expenditures, monthly (yearly) unrevised 328

Classi�ed re eipts and expenditures, monthly

1938

unrevised 381/389

Classi�ed re eipts and expenditures, monthly (yearly) unrevised 389

Classi�ed re eipts and expenditures, monthly

1939

unrevised 339/347

Classi�ed re eipts and expenditures, monthly (yearly) unrevised 347

Classi�ed re eipts and expenditures, monthly

1940

unrevised 614/621

Classi�ed re eipts and expenditures, monthly (yearly) unrevised 621

Table 8: Table names and page numbers from the Annual Reports of the Se retary of the

Treasury for interest expenditures

44

Ja obson, Leeper, & Preston: 1933

A.1.3 Cal ulating Monthly Net Interest Be ause interest re eipts are only avail-

able on a yearly basis, we are only able to al ulate net interest on a yearly basis. We then

use the yearly net interest series to impute monthly net interest. We �rst al ulate the ratio

of yearly interest re eipts to yearly interest expenditures and then multiply this ratio by

monthly interest expenditures to impute monthly interest re eipts. Let the expression for

imputed interest re eipts in month t be given as:

Imputed Monthly Interest Re eiptst =Yearly Interest Re eipts

Yearly Interest Expenditures

∗Monthly Interest Expenditurest

Monthly net interest is then al ulated as:

Imputed Monthly Net Interestt = Monthly Interest Expenditurest−Imputed Monthly Interest Re eiptst

A.2 Federal Re eipts and Expenditures

This se tion details how our series of monthly federal re eipts and expenditures from July

1919 to June 1940 from the Annual Reports of the Se retary of the Treasury on the State

of Finan es di�er from other sour es. We use data for re eipts and expenditures that was

revised in 1933 to � over all expenditures of the Re onstru tion Finan e Corporation, in-

luding payments against redits established for the orporation through the pur hase of its

notes under se tion 9 of the Re onstru tion Finan e Corporation A t.�

36

We use data on an

unrevised ash basis for re eipts and expenditures.

37

Our three main sour es of omparison are data from the NBER Ma ro History Database

(NBER)

38

, Firestone's (1960) book, and Romer (1992) who uses re eipts and outlays

39

from

the 1979 Statisti al Appendix to the Annual Report, table 2, pp. 4-11 [Romer (1992)℄.

A.2.1 Federal Re eipts Re eipts from Firestone orrespond to our series ex ept for

�s al years 1931, 1932, and 1940. On page 80, Firestone explains that trust fund re eipts

were eliminated from internal revenue after June 1932 and his series take into a a ount this

revision ba k to July 1930. Firestone (page 82) also dedu ts net transfers from the Federal

Old-Age and Survivors Insuran e Trust Fund from re eipts to obtain lower monthly re eipts

for �s al year 1940. The NBER re eipts data is split into three re eipt series a, b, and .

NBERa mat hes our series up to �s al year 1932. NBERb mat hes Firestone for �s al years

1931 and 1932 � also taking into a ount the elimination of trust fund re eipts � and then

tra ks our series through �s al year 1940. NBER (not shown) also dedu ts net transfers

from the Federal Old-Age and Survivors Insuran e Fund and thus tra ks Firestone for �s al

year 1940.

36

Footnote 1, Table 6, page 312 of Annual Report of the Se retary of the Treasury on the State of the

Finan es for Fis al year ended June 30, 1933

37

See footnote 33 for an explanation of a ounting onventions.

38

A essed via the NBER's Ma rohistory Database, Chapter 15

39

Starting in 1968, the Department of the Treasury (various) introdu ed new uni�ed budget on epts

in luding outlays. On page 8, the report explains that federal outlays in lude loans and expenditures.

45

Ja obson, Leeper, & Preston: 1933

2

4

6

8

10

1920 1925 1930 1935 1940

Receipts Total expenditures

Figure 23: Fis al Year Totals of Monthly Re eipts and Total Expenditures, billions of dollars.

Sour e: Department of the Treasury (various). See Table 9 for details.

2

3

4

5

6

7

1920 1925 1930 1935 1940

Our series Firestone

2

3

4

5

6

7

1920 1925 1930 1935 1940

Our series NBERa NBERb

Figure 24: Fis al year totals of monthly re eipts and total expenditures, billions of dollars.

Sour e: Department of the Treasury (various) (see Table 9 for details); Firestone (1960);

NBER Ma rohistory database (m15004b,m15004 ).

Our yearly totals of monthly re eipts data do not always mat h the yearly totals in other

tables in the annual reports. Although the yearly data is revised throughout various annual

reports, the monthly is not. The yearly re eipts data is unrevised from �s al years 1920 to

46

Ja obson, Leeper, & Preston: 1933

1935. In 1936, the data is revised starting in 1931. Our series of annual totals of monthly

re eipts data mat hes the yearly data until �s al year 1933 when our series turns slightly

lower.

2.0

3.0

4.0

5.0

6.0

7.0

1920 1925 1930 1935

Our series 1935

Figure 25: Fis al year totals of monthly re eipts and re eipts by �s al year, billions of dollars.

Sour e: Department of the Treasury (various). See Table 9 for details.

Annual re eipts data remains unrevised from �s al years 1936 to 1939. In 1939, re-

eipts were mostly revised downwards for �s al years 1931 through 1935. This revised series

mat hes our series from �s al years 1933 through 1939. In 1940, re eipts data was revised

downwards for �s al years 1937 through 1940.

40

2.0

3.0

4.0

5.0

6.0

7.0

1920 1925 1930 1935 1940

Our series 1939 1940

Figure 26: Fis al year totals of monthly re eipts and re eipts by �s al year, billions of dollars.

Sour e: Department of the Treasury (various). See Table 9 for details.

40

Footnote 14 on Page 649 of the 1940 Annual Report explains that: �In the �s al year 1941 amounts rep-

resenting appropriations equal to `So ial Se urity-Unemployment taxes' olle ted and deposited as provided

under se . 201 (a) of the So ial Se urity A t Amendments of 1939, less reimbursements to the General Fund

for administrative expenses, are dedu ted on the daily Treasury statement from total re eipts. Su h net

amounts are re�e ted under trust a ount re eipts as net appropriations to the Federal old-age and survivors

insuran e trust fund. The �s al years 1937, 1938, and 1939, have been revised in this statement to re�e t

similar treatment. Fis al year 1940 �gures are also on this revised basis.�

47

Ja obson, Leeper, & Preston: 1933

A.2.2 Federal Expenditures Firestone and the NBER use ordinary expenditures for

their expenditure series starting in De ember 1920 through �s al year 1933 (June 1933).

Romer uses ordinary outlays through �s al year 1933.

41

Ordinary expenditures are a subset

of total expenditures and ex lude publi debt retirements. For �s al years 1920 through 1926,

ordinary expenditures ex lude pur hases of obligations of foreign governments in addition to

publi debt retirements. Starting in �s al year 1934, the Annual Report of the Se retary of

the Treasury divides total expenditures into general and emergen y ategories.

42

Starting in

1934, Firestone, the NBER, and Romer begin using total expenditures for their expenditures

series. We use total expenditures throughout the entire sample. Prior to �s al year 1934,

total expenditures are on average roughly 13 per ent higher than ordinary expenditures.

The expenditure series from Firestone mat hes our series of ordinary expenditures from

1922 through �s al year 1930. Firestone explains on page 82 that starting in �s al year

1931, trust fund transa tions were eliminated from ordinary expenditures hargeable against

ordinary re eipts. Trust fund expenditures were, however, still in luded in ordinary re eipts

through 1933 for omparison purposes. Our yearly totals of monthly ordinary expenditures

diverge from Firestone's from �s al years 1931 to 1933. Firestone's data for January 1932 to

June 1933 mat hes that of NBER (not shown). Our series of ordinary expenditures mat hes

NBERb up to �s al year 1933. Romer's series of ordinary outlays is almost always lower

than our series and those given by the NBER and Firestone.

3.0

4.0

5.0

6.0

7.0

1920 1925 1930 1935

Our series, ord. Firestone, ord. Our series, total Romer

3.0

4.0

5.0

6.0

7.0

1920 1925 1930 1935

Our series, ord. NBERb, ord. Our series, total Romer

Figure 27: Fis al year totals of monthly ordinary expenditures, billions of dollars. Sour e:

Department of the Treasury (various) (see Table 9 for details); Firestone (1960); NBER

Ma rohistory database (m15004b,m15004 ).

The total expenditure series from Firestone mat hes NBER from �s al year 1934 through

�s al year 1937. From �s al year 1937 through 1939, Firestone's data mat hes NBERd.

Firestone explains on page 84 that under an a t of February 1938, the Se retary of the

41

See footnote 39 for the di�eren e between outlays and expenditures.

42

Table 6 Footnote 6 on page 316 from the Annual Report of the Se retary of the Treasury on the State

of the Finan es for Fis al year ended June 30, 1934 explains that �Emergen y expenditures prior to the

�s al year 1934 (ex ept Re onstru tion Finan e Corporation) are in luded in general expenditures, the

lassi� ation of whi h emergen y expenditures is not available for omparison with emergen y expenditures

for the �s al year 1934. Therefore, neither the totals of general expenditures nor the totals of emergen y

�s al expenditures for the �s al year 1934 are omparable with the total of prior �s al years.�

48

Ja obson, Leeper, & Preston: 1933

Treasury an eled $2.7 billion of obligations pur hased from the RFC whi h the RFC ould

not repay to the Treasury. As a onsequen e, budget expenditures show only amounts spent

from funds allo ated by the RFC for purposes for whi h no provisions for repayment to the

Treasury were made.The series from Firestone mat hes NBERe (not shown) for �s al year

1940. Our series is larger than Firestone's and NBER from 1934 through 1938. Although

the gap shrinks from 1938 through 1940, our series is slightly higher than the other three

series. Romer's series of total outlays is below our series and those given by the NBER and

Firestone for most years.

6.0

7.0

8.0

9.0

10.0

1934 1936 1938 1940

Firestone, total NBERc, total Our series, total Romer

6.0

7.0

8.0

9.0

10.0

1934 1936 1938 1940

Firestone, total NBERd, total Our series, total Romer

Figure 28: Fis al Year Totals of Monthly Total Expenditures, billions of dollars. Sour e:

Department of the Treasury (various) (see Table 9 for details); Firestone (1960); NBER

Ma rohistory database (m15004b,m15004 ).

As with the re eipts series, our series for total and ordinary expenditure do not always

mat h yearly data given elsewhere in the annual reports. From �s al year 1922 to �s al

year 1931 our series of yearly totals of monthly expenditures data mat h yearly totals given

elsewhere in the annual reports on an unrevised ash basis. In the 1927 annual report,

ordinary expenditures are revised upwards. In the 1933 annual report, total and ordinary

expenditures are revised for �s al years 1932 and 1933. These revisions di�er from revisions

overing the expenditures of the Re onstru tion Finan e Corporation in 1933.

49

Ja obson, Leeper, & Preston: 1933

3

4

5

6

1920 1925 1930 1935

Our series, ordinary 1933, ordinary

3

4

5

6

1920 1925 1930 1935

Our series, total 1933, total

Figure 29: Fis al year totals of monthly ordinary and total expenditures and ordinary and

total expenditures by �s al year, billions of dollars. Sour e: Department of the Treasury

(various) (see Table 9 for details).

As mentioned previously, starting in 1934 until 1939, monthly expenditures are split into

general and emergen y expenditures ategories rather than ordinary and total expenditures

ategories. Tables of yearly totals ontinue to ategorize expenditures into ordinary and

total even though the monthly series does not maintain this distin tion. Our yearly totals of

monthly ordinary expenditures stop in 1934 and we instead ompute general expenditures

for 1934-1939. Yearly ordinary and total expenditure series in the table are not revised from

1933 to 1935. Starting in 1936, the yearly ordinary and total expenditure series are revised

ba k to 1930. Our series of total expenditures is lower than the 1935 and 1936 yearly series.

2

4

6

8

1920 1925 1930 1935

Our series, ordinary Our series, general

1935, ordinary 1936, ordinary

4

6

8

10

1920 1925 1930 1935

Our series, total 1935, total 1936, total

Figure 30: Fis al year totals of monthly ordinary and total expenditures and ordinary and

total expenditures by �s al year, billions of dollars. Sour e: Department of the Treasury

(various) (see Table 9 for details).

Yearly ordinary and total expenditures are revised in 1937, 1939, and 1940. The 1937

total expenditure series mat hes our series of yearly totals of monthly data the best.

50

Ja obson, Leeper, & Preston: 1933

2

4

6

8

10

1920 1925 1930 1935 1940

Our series, ord. Our series, general 1937, ord.

1939, ord. 1940, ord.

4

6

8

10

1920 1925 1930 1935 1940

Our series, total 1937, total

1939, total 1940, total

Figure 31: Fis al year totals of monthly total expenditures, billions of dollars. Sour e:

Department of the Treasury (various) (see Table 9 for details).

51

Ja obson, Leeper, & Preston: 1933

Table name Year Re eipts page Expenditures page

Ordinary re eipts, and expenditures hargeable against (monthly)

1920

see 1921

Ordinary re eipts, and expenditures hargeable against (yearly) see 1922

STATEMENT SHOWING CLASSIFIED RECEIPTS...

1921

240 241

Ordinary re eipts, and expenditures hargeable against (yearly) see 1922

Ordinary re eipts, and expenditures hargeable against (monthly)

1922

270 271

Ordinary re eipts, and expenditures hargeable against (yearly) 270 271

Ordinary re eipts, and expenditures hargeable against (monthly)

1923

512 513

Re eipts and expenditures of the United States Government... 512 513

Ordinary re eipts, and expenditures hargeable against (monthly)

1924

378 379

Re eipts and expenditures of the United States Government... 378 379

Ordinary re eipts, and expenditures hargeable against (monthly)

1925

472 474

Re eipts and expenditures of the United States Government... 472 474

Ordinary re eipts, and expenditures hargeable against (monthly)

1926

445 447

Ordinary re eipts, and expenditures hargeable against (yearly) 443 443

Ordinary re eipts, and expenditures hargeable against (monthly)

1927

462 462

Ordinary re eipts, and expenditures hargeable against (yearly) 445 445

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1928

424 424

Summary of ordinary re eipts, expenditures hargeable (yearly)... 407 407

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1929

412 412

Summary of ordinary re eipts, expenditures hargeable (yearly)... 394 394

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1930

506 506

Summary of ordinary re eipts, expenditures hargeable (yearly).... 488 488

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1931

462 462

Ordinary re eipts, expenditures hargeable against (yearly)... 448 448

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1932

370 370

Re eipts and expenditures for the �s al years 1789 to... 365 369

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1933

312 312

Re eipts and expenditures for the �s al years 1789 to... 306 310

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1934

306 306

Re eipts and expenditures for the �s al years 1789 to... 301 305

Summary of ordinary re eipts, expenditures hargeable (monthly)...

1935

328 328

Re eipts and expenditures for the �s al years 1789 to... 323 327

Classi�ed re eipts and expenditures, monthly...

1936

337 339/342

Re eipts and expenditures for the �s al years 1789 to... 359 363

Classi�ed re eipts and expenditures, monthly...

1937

320 322/324

Re eipts and expenditures for the �s al years 1789 to... 349 353

Expenditures by major fun tions for the �s al years 1930-1937 354

Classi�ed re eipts and expenditures, monthly...

1938

379 381/384

Re eipts and expenditures for the �s al years 1789 to... 413 417

Expenditures by major fun tions for the �s al years 1931-1938 418

Classi�ed re eipts and expenditures, monthly...

1939

337 339/342

Re eipts and expenditures for the �s al years 1789 to... 361 365

Expenditures by major fun tions for the �s al years 1931-1939 367

Classi�ed re eipts and expenditures, monthly...

1940

612 615/616

Re eipts and expenditures for the �s al years 1789 to... 645 649

Expenditures by major fun tions for the �s al years 1933-1940 653

Re eipts in general and spe ial a ounts, by major sour es... 651

Table 9: Table names and page numbers from the Annual Reports of the Se retary of the

Treasury for federal re eipts and expenditures

52

Ja obson, Leeper, & Preston: 1933

A.3 Market Value and Returns

The following se tion details our al ulation of market value and return on the United States's

bond portfolio. We use data from Hall and Sargent (2015), provided to us by the authors, as

well as the CRSP to obtain the quantity, pri e, a rued interest, interest rate, and oupon

frequen y of ea h government se urity outstanding in a given month.

Let Bit(t+ j) denote the dollar value of type i bonds outstanding in period t that mature

in period t + j and QDit (t + j) be the dirty pri e (pri e+a rued interest) of su h bonds.

Be ause the number of types of bonds of a ertain maturity ea h period an vary over time,

we let Nt(t + j) represent the number of su h bonds in period t.Let Bt(t+ j) denote the dollar value of all bonds outstanding in period t that mature in

period t + j, de�ned as

Bt(t+ j) =

Nt(t+j)∑

i=1

Bit(t + j) (23)

Then the par value of all debt outstanding at the end of period t�the fa e value of the bond

portfolio�is

BMt =

∞∑

j=1

Nt(t+j)∑

i=1

Bit(t+ j) =∞∑

j=1

Bt(t+ j) (24)

De�ne νi(t+ j) as the share of se urity of type i that is outstanding at t and matures at

t+ j

νi(t+ j) =Bit(t+ j)

∑Nt(t+j)i=1 Bit(t + j)

=Bit(t + j)

Bt(t + j)(25)

where

∑Nt(t+j)i=1 νi(t + j) = 1. Then the weighted dirty pri e of bonds outstanding at t that

mature in t+ j is

QDt (t+ j) = Qt(t+ j)+AIt(t+ j) =

Nt(t+j)∑

i=1

(

Qit(t+ j) + AIit(t+ j))

νi(t+ j) (26)

where Qt(t+ j) is the lean pri e of bonds outstanding at t that mature in t+ j, AIt(t+ j) isthe a rued interest on bonds outstanding at t that mature in t+ j. For zero- oupon bonds,

the dirty pri e is equal to the lean pri e.

We also de�ne µt(t + j) as the share of the total par value of bonds outstanding at the

end of t that matures in t+ j

µt(t+ j) =Bt(t+ j)

BMt

(27)

where

∑∞

j=1 µt(t+ j) = 1. This permits us to de�ne the nominal pri e of the bond portfolio,

PMt , as

PMt =

∞∑

j=1

QDt (t+ j)µt(t + j) (28)

53

Ja obson, Leeper, & Preston: 1933

With a omplete and general maturity stru ture, the government's budget identity is

∞∑

j=0

(QD

t (t+ j) + IPt(t+ j))Bt−1(t + j) = Ptst +

∞∑

j=1

QDt (t+ j)Bt(t+ j) (29)

Where QDt (t) ≡ 1 and IPt(t + j) is the interest payable on bonds outstanding at t that

mature in t + j. Interest payable is an government expense in period t and is thus in luded

in the government budget identity.

The market value of debt outstanding in period t is

PMt BM

t ≡

∞∑

j=1

QDt (t+ j)Bt(t+ j) (30)

so that the omparable expression at t− 1 is

PMt−1B

Mt−1 ≡

∞∑

j=1

QDt−1

((t−1)+(j+1)

)Bt−1

((t−1)+(j+1)

)=

∞∑

j=1

QDt−1(t+j)Bt−1(t+j) (31)

The arry-over market value uses the same bonds as the market value for period t − 1but using period t dirty pri es and intermediate oupon payments. The arry-over pri e,

PCt , re�e ts oupon payments that were paid between periods t − 1 and t. The arry-over

market value is de�ned as

PCt BM

t−1 ≡

∞∑

j=0

(

QDt (t+ j) + IPt(t + j)

)

Bt−1(t+ j) (32)

IPt(t+ j) is the interest payable on bonds outstanding at t that mature in t+ j. PCt di�ers

from its dirty-pri e analog only when there is a oupon payment in month t. Figure 32

illustrates the timing of oupon payments.

PMt−1B

Mt−1

t− 1

PCt BM

t−1 PMt BM

t

t

PCt+1B

Mt

Figure 32: Timing of a tual and arry-over market value

Using the de�nitions of market value and arry over market value, (29) an be written

as:

PCt BM

t−1 = Ptst + PMt BM

t (33)

Multiplying and dividing the left hand side by last period's market value allow the govern-

ment budget identity to be expressed in terms of the rate of return on government debt:

PCt BM

t−1

PMt−1B

Mt−1

︸ ︷︷ ︸

rate of return

PMt−1B

Mt−1 = Ptst + PM

t BMt (34)

54

Ja obson, Leeper, & Preston: 1933

The rate of return an also be derived by de omposing hanges in market value into rates

of return and hanges in size. We start by expanding the ratio of period t to period t − 1market value

PMt BM

t

PMt−1B

Mt−1

≡PCt BM

t−1

PMt−1B

Mt−1

︸ ︷︷ ︸

rate of return

·PMt BM

t

PCt BM

t−1︸ ︷︷ ︸

size ratio

(35)

The expression for the rate of return is the same as (34) and an be expressed as

PCt BM

t−1

PMt−1B

Mt−1

=

∑∞

j=1

(

Qt(t + j) + AIt(t+ j) + IPt(t + j))

Bt−1(t+ j)

∑∞

j=1

(

Qt−1(t+ j) + AIt−1(t+ j))

Bt−1(t+ j)(36)

This rate of return re�e ts the per entage hange in the value of the bond portfolio between

period t− 1 and t, holding the bond portfolio �xed.

The size ratio an be expressed as

PMt BM

t

PCt BM

t−1

=

∑∞

j=1

(

Qt(t + j) + AIt(t + j))

Bt(t+ j)

∑∞

j=1

(

Qt(t+ j) + AIt(t+ j) + IPt(t+ j))

Bt−1(t + j)(37)

Changes in size in orporates new issues, redemptions, and oupon payments that o ur

between periods t− 1 and t. The size ratio re�e ts the per entage hange in the value of the

bond portfolio that arises from hanges in the bond portfolio itself, in luding any hanges in

maturity stru ture.

rMt =PCt BM

t−1/Pt

PMt−1B

Mt−1/Pt−1

=

∑∞

j=1Qt(t+ j)Bt−1(t+ j)/Pt∑

∞

j=1Qt−1(t+ j)Bt−1(t+ j)/Pt−1

(38)

Of ourse, the identity (34) an be expressed in real terms as:

rMt PMt−1b

Mt−1 = st + PM

t bMt (39)

where bMt ≡ BMt /Pt is the real par value of debt outstanding at t.

The surprise omponent in the real return on the bonds portfolio is:

ηDt ≡ rMt −Et−1rMt (40)

Using Et−1[QDt (t+j)/Pt] =

(Qt−1(t+j)+AIt(t+j)+IPt(t+j)

)/Pt−1, then the expe tation

is of no real apital gain or loss on the portfolio. A rued interest, AIt(t + j), and interest

payable, IPt(t + j), of bonds outstanding in period t that mature in period t + j is known

in period t− 1. Hen e, Et−1[AIt(t+ j) + IPt(t+ j)] = AIt(t+ j) + IPt(t+ j). The surprisein the real return be omes

ηDt =

∞∑

j=0

( (Qt(t+ j) + AIt(t+ j) + IPt(t+ j)

)/Pt

(Qt−1(t+ j) + AIt(t+ j) + IPt(t+ j)

)/Pt−1

− 1

) (Qt−1(t+ j) +AIt(t+ j) + IPt(t+ j)

)Bt−1(t+ j)

PMt−1

BMt−1

(41)

55

Ja obson, Leeper, & Preston: 1933

Real returns an be s aled by omponents isolating hanges in the pri e level and hanges

in bond pri es. Re-writing (41) as:

ηDt =P ct B

Mt−1/Pt

PMt−1B

Mt−1/Pt−1

︸ ︷︷ ︸

rDt

−P ct Bt−1

PMt−1Bt−1︸ ︷︷ ︸

RDt

+

∑∞

j=1

(Qt(t+ j)−Qt−1(t+ j)

)Bt−1(t+ j)

PMt−1B

Mt−1

(42)

Whi h an be further re-arranged to:

ηDt = RDt (1/πt − 1)︸ ︷︷ ︸

due to pri e level

+RDt

(∑∞

j=1

(Qt(t + j)−Qt−1(t+ j)

)Bt−1(t + j)

PCt BM

t−1

)

︸ ︷︷ ︸

due to bond pri es

(43)

If there are no hanges in the pri e level between periods t − 1 and t, i.e. π = 1 and

weighted hanges in bond pri es sum to zero

∑∞

j=1Qt(t+ j)−Qt−1(t+ j) = 0 , then ηDt = 0indi ating no apital gains or losses. If there is no hange in the pri e level (πt = 1) thenRD

t (1/πt − 1) = 0 then apital gains or losses an be interpreted as the weighted hange in

bond pri es as a share of market value s aled by nominal returns. If the weighted hanges

in bond pri es sum to zero,

(∑∞

j=1

(Qt(t+ j)−Qt−1(t+ j)

)= 0, then apital gains or losses

are hanges in the pri e level s aled by nominal returns.

Real and nominal returns are denominated in per entage points of market value out-

standing at Bt−1

Abandon gold standard0.96

0.98

1.00

1.02

1.04

1926 1928 1930 1932 1934 1936 1938 1940

Real return Nominal return

Figure 33: Real and nominal pri e returns

Real returns to U.S. debt show a mu h larger drop than nominal returns to U.S. debt

after the departure form the gold standard.

56

Ja obson, Leeper, & Preston: 1933

Abandon gold standard

-0.04

-0.02

0.00

0.02

0.04

1926 1928 1930 1932 1934 1936 1938 1940

dirty prices clean prices

Figure 34: Real innovations to pri e returns with lean and dirty pri es

Innovations show large losses after the abandonment of the gold standard.

Innovations apture the unexpe ted losses or gains on U.S. debt due to bond pri es or the

pri e level. We multiplying innovations by the beginning of period market value (PMt−1B

Mt−1)

to apture the dollar amount of the di�eren e between real and expe ted real returns to

holding U.S. debt. We then take this dollar amount as ratio of the urrent period market

value (PMt BM

t ) to apture surprise apital gains or losses as a per ent of market value. Figure

35 is thus:

ηDtPMt−1B

Mt−1

PMt BM

t

∗ 100 (44)

Abandon gold standard−4.00

−2.00

0.00

2.00

4.00

1926 1928 1930 1932 1934 1936 1938 1940

Figure 35: Capital gains and loss as a per ent of market value (44)

57

Ja obson, Leeper, & Preston: 1933

Abandon gold standard

−0.03

−0.02

−0.01

0.00

0.01

0.02

1926 1928 1930 1932 1934 1936 1938 1940

Innovation due to price level Innovation due to bond prices

Figure 36: Innovations to pri e returns de omposed into hanges from bond pri es and

hanges from the pri e level (43)

After the abandonment of the gold standard, the pri e level is largely responsible for the

apital loss on holding government debt.

B The Appendix

The following parametri assumptions are made se tion ??. Begin with the government's

budget onstraint in steady state

P lBl

P

(1− β−1

)= F − T.

Then

P lBl

PY

(β−1 − 1

)=

T − F

Yimplies an assumption on the steady state debt to GDP ratio pins down the stru tural

surplus. Assume an annual debt-to-GDP ratio of 30 per ent. This implies

P lBl

PY= 1.2

in a quarterly model. Assuming

C

Y= 0.8 and

F

T= 0.2

determines the tax to GDP ratio residually. In turn an assumption on the fra tion of gov-

ernment spending in output determines steady state taxation. Furthermore

PT

Bl=

T

Y

Y P

Bl

PF

Bl=

F

Y

Y P

Bl

58

Ja obson, Leeper, & Preston: 1933

where the right hand sides of ea h expression are already determined ratios. To determine

the other ratios in the government budget onstraint alibrate

M

P lBl= 1

whi h orresponds to the ratio of M1 to the market value of debt in 1933. This permits

P gGm

P lBl=

P gGm

M

M

P lBl= α

M

P lBl

whi h ompletes the solution for required ratios.

Other parameter values whi h are pi ked fairly arbitrarily: β = 0.99, σ = 1, ϕ = 20,κ = 100, α = 0.4, ρ = 0.95. The sho ks all have auto regressive oe� ient 0.5. From

the liquidity preferen e s hedule, (??),the elasti ity of money demand with respe t to the

interest rate is

β

(1− β)ϕ.

For values of this elasti ity around unity, the parameter ϕ must be of the order of 100.The basi patterns observed in the impulse responses don't depend mu h on the assumed

alibration. Poli y parameters are given by: γb = 0.1 under the gold standard. In the

unba ked �s al expansion γb = 0 and φπ = 0.9.

C Additional VAR Results

This appendix reports a more omplete set of VAR results than those in the text. Figure

37 reports a tual data and un onditional fore asts for the seven series in the VAR. Figure

38 shows the full moving average representations for the seven-variable VAR estimated over

the unba ked �s al expansion period (April 1933 to June 1940).

59

Ja obson,Leeper,&Preston:1933

1934 1935 1936 1937 1938 1939 1940

0.8

1

1.2

1.4R Actual

R Forecast

1934 1935 1936 1937 1938 1939 1940

10

15

20M Actual

M Forecast

1934 1935 1936 1937 1938 1939 1940

-1.5

-1

-0.5

S Actual

S Forecast

1934 1935 1936 1937 1938 1939 1940

25

30

35

40

45

B Actual

B Forecast

1934 1935 1936 1937 1938 1939 1940

5

10

15

G Actual

G Forecast

1934 1935 1936 1937 1938 1939 1940

80

82

84

86

88

P Actual

P Forecast

1934 1935 1936 1937 1938 1939 1940

80

90

100

110

Y Actual

Y Forecast

µ = 0.6 0.3 1 1.75 2 2

Figure 37: A tual and un onditional fore asts of variables in VAR using the hyperparameters λ0 = 0.6, λ1 = 0.3, λ3 = 1.0, λ4 =1.75, µ5 = µ6 = 2.0, in the notation of Sims and Zha (1998).

60

Ja obson,Leeper,&Preston:1933

-3.6561

0

4.8553

R

10-4 MP

MD

FP

B

G

P

Y

-0.0205

0

0.0270

M

-0.2363

00.0364

S

-6.47200

13.0448

B

10-3

-0.02660

0.0596

G

-9.2751

0

7.6151

P

10-3

6 18 30-0.0108

0

0.0255

Y

6 18 30

6 18 30

6 18 30

6 18 30

6 18 30

6 18 30

Figure 38: Full moving average representation of the identi�ed VAR estimated over the unba ked �s al expansion period (April

1933 to June 1940). Solid lines are maximum likelihood estimates; dashed lines are 68 per entile probability bands based on

1000 draws from the posterior distribution of all the VAR parameters.

61

Ja obson, Leeper, & Preston: 1933

D Fis al Impli ations of Gold Sterilization

D.1 Gold Sterilization during the Great Depression

Gold imports have the potential to in rease the monetary base of an e onomy following

the lassi al gold standard or the gold ex hange standard. Poli ymakers an ountera t the

in rease in the monetary base by sterilizing gold in�ows whi h entails paying for imported

gold in government se urities rather than bank reserves. Prior to 1933, the Federal Reserve

ondu ted gold import operations and sterilization de isions. By June of 1934, these re-

sponsibilities shifted to the Treasury as the result of a series of presidential pro lamations,

exe utive orders, joint-resolutions, and A ts that ulminated in an embargo on gold exports

43

and the Treasury seizing the entire monetary gold sto k in luding oins and bullion held by

private itizens, business, and the Federal Reserve Banks.

44

Massive gold imports more than tripled the monetary gold sto k from $4.25 billion at

the start of 1933 to $14.42 billion at the end of 1938. Meltzer (2003, p. 459) notes that the

Treasury pur hased more than $4 billion of gold from 1934-1936. Friedman and S hwartz

(1963, p. 545) attribute the gold in�ows throughout this period to the depre iation of the

dollar, Hitler's rise to power, and the outbreak of war in Europe. Studenski and Krooss

(1952, p. 394) in lude the Treasury's $35 an oun e pur hase pri e for gold, favorable trade

balan es, and the reditor position of the United States as additional fa tors that in reased

gold imports. To our knowledge, the Gold Reserve A t of 1934's ban on private itizens

holding monetary gold required banks to sell newly imported gold to the Treasury.

45

With

gold in�ows pushing up ex ess reserves, poli ymakers feared that the growing monetary base

ould ignite in�ationary for es [Jaremski and Mathy (2016)℄. To urb the growth of ex ess

reserves and hen e the monetary base, the Treasury sterilized gold imports from De ember

1936 to April 1938.

Expanding on the example provided by Johnson (1939, p. 144), we illustrate the e�e ts

of the Treasury's non-sterilized and sterilized gold pur hases on the balan e sheets of the

Treasury, the Federal Reserve, and member banks.

43

Exe utive Order 6111 on Transa tions in Foreign Ex hange was implemented on April 20, 1933. See

http://www.presiden y.u sb.edu/ws/index.php?pid=14621

44

See Bordo, Humpage, and S hwartz (2015, pp. 56�57) for a detailed time line of events. Jaremski and

Mathy (2016, p. 6) report that most gold imports ame through New York City's gold market and New

York City banks ontinued to sell their gold to the Federal Reserve Bank of New York who a ted as �s al

agent to the Treasury, the ultimate pur haser of the gold.

45

Bordo, Humpage, and S hwartz (2015, p. 65) explain that the Treasury issued spe ial li enses for

ommer ial banks to obtain gold for ustomers. This suggests that banks were not allowed to keep gold on

their balan e sheets.

62

Ja obson, Leeper, & Preston: 1933

1. Gold Imported by Member Banks Member banks import $1,000 worth of gold

and fund it by issuing $1,000 worth of deposits. Member bank assets and liabilities rise by

$1,000.

Treasury

Assets Liabilities

Federal Reserve

Assets Liabilities

Member Banks

Assets Liabilities

+$1,000 gold +$1,000 deposits

+$1,000 +$1,000

2. High Powered Money Creation Member banks sell their imported gold to the

Federal Reserve for $1,000. The Federal Reserve pays for the gold by issuing reserves to

member banks whi h in reases high-powered money by $1,000. For member banks, gold

is swapped for reserves and their aggregate asset position is un hanged �both assets and

liabilities remain elevated by the original $1,000 inje tion. If the Federal Reserve did not want

to sterilize and they were responsible for sterilization de ision, this would be the �nal step.

Skip to step 2b at the end of this Appendix for the e�e ts of Federal Reserve sterilization.

Treasury

Assets Liabilities

Federal Reserve

Assets Liabilities

+$1,000 gold +$1,000

reserves

+$1,000 +$1,000

Member Banks

Assets Liabilities

-$1,000 gold $1,000 deposits

+$1,000 reserves

$1,000 $1,000

3. Gold Transferred to Treasury Under the Gold A t of 1934, gold ould not be

exported and any imported gold had to be turned over to the Treasury. As noted by Jaremski

and Mathy (2016, p. 6), the Federal Reserve would then transfer the gold to the Treasury who

paid for the gold by drafting on its balan es at the Federal Reserve. Although the aggregate

value of the balan e sheets of the Treasury and the Federal Reserve are un hanged, the

omposition of their balan e sheets hange.

Treasury

Assets Liabilities

+$1,000 gold

-$1,000 due

from Fed

Federal Reserve

Assets Liabilities

-$1,000 gold $1,000 reserves

-$1,000 due

to Treasury

Member Banks

Assets Liabilities

$1,000 reserves $1,000 deposits

$1,000 $1,000

63

Ja obson, Leeper, & Preston: 1933

4a. No Sterilization Under the Treasury: The Treasury replenishes its balan es at

the Federal Reserve by issuing gold erti� ates and depositing them at the Federal Reserve

as the �nal payment for gold pur hases. Non-sterilized gold imports ultimately in rease the

balan e sheets of the Treasury, the Federal Reserve, and member banks and leave the amount

of free-gold at the Treasury un hanged. The Treasury does not o�set the reation of high

powered money by retiring the newly reated reserves as will be the ase with sterilization.

Importantly, in the ase of no sterilization, there is no in rease in Treasury indebtedness

to the private se tor be ause the Treasury reates �money� through gold erti� ates.

Treasury

Assets Liabilities

+$1,000 due +$1,000 gold

from Fed erti� ates to

Treasury

+$1,000 +$1,000

Federal Reserve

Assets Liabilities

+$1,000 gold $1,000 reserves

erti� ates

from Treasury

+$1,000 +$1,000

Member Banks

Assets Liabilities

$1,000 reserves $1,000 deposits

$1,000 $1,000

4b. Sterilization Under the Treasury: When sterilizing gold imports, the Treasury

replenishes balan es at the Federal Reserve by selling government se urities to member banks

rather than issuing gold erti� ates and depositing them at the Federal Reserve. The Federal

Reserve again settles the transa tion between the Treasury and member banks through

reserves. Member banks pay for se urity sales by retiring reserves outstanding at the Federal

Reserve. The Federal Reserve then o�sets this transa tion by rediting their balan e due

to the Treasury/debiting the Treasury's balan es held at the Federal Reserve. Sterilization

in reases the aggregate balan e sheets of the Treasury and member banks, but not the

Federal Reserve.

In this ase, there is an in rease in Treasury indebtedness to the private se tor and there

is no in rease in bank reserves.

Treasury

Assets Liabilities

+$1,000 due +$1,000 gov't

from Fed se urities

+$1,000 +$1,000

Federal Reserve

Assets Liabilities

-$1,000 reserves

+$1,000 due to

Treasury

Member Banks

Assets Liabilities

-$1,000 reserves $1,000 deposits

+$1,000 gov't

se urities pur hased

from Treasury

$1,000 $1,000

64

Ja obson, Leeper, & Preston: 1933

2b. Sterilization Under the Federal Reserve When sterilizing gold imports, the

Federal Reserve pays for gold by selling government se urities to member banks rather than

reating reserves as seen in step 2. Sterilization leaves the aggregate balan e sheets of the

Federal Reserve and the Treasury un hanged while the balan e sheet of member banks is

expanded. In the ase of Federal Reserve sterilization, there is no in rease in Treasury

indebtedness. Be ause se urity sales by the Federal Reserve prevent the reation of reserves,

sterilization by the Federal Reserve is equivalent to ontra tionary open market operations.

Treasury

Assets Liabilities

Federal Reserve

Assets Liabilities

+$1000 gold

-$1000 gov't

se urities

Member Banks

Assets Liabilities

-$1000 gold $1000 deposits

+$1000 gov't

se urities

$1000 $1000

65

Ja obson, Leeper, & Preston: 1933

Referen es

Balke, N., and R. J. Gordon (1986): �Appendix B: Histori al Data,� in The Ameri an

Business Cy le: Continuity and Change, ed. by R. J. Gordon, pp. 781�850. University of

Chi ago Press, Chi ago.

Barro, R. J. (1979): �Money and the Pri e Level Under the Gold Standard,� The E onomi

Journal, 89(353), 13�33.

Beard, T. R., and W. D. M Millin (1991): �The Impa t of Budget De� its in the

Interwar Period,� Journal of Ma roe onomi s, 13(2), 239�266.

Blum, J. M. (1959): From the Morgenthau Diaries: Years of Crisis, 1928�1938. Houghton

Mi�in Company, Boston.

Board of Governors of the Federal Reserve System (1937): �23rd Annual Re-

port,� .

Bordo, M. D., O. F. Humpage, and A. J. S hwartz (2015): Strained Relations: U.S.

Foreign Ex hange Operations and Monetary Poli y in the Twentieth Century. University

of Chi ago Press, Chi ago.

Bordo, M. D., and F. E. Kydland (1995): �The Gold Standard As a Rule: An Essay in

Exploration,� Explorations in E onomi History, 32(4), 423�464.

Brown, E. C. (1956): �Fis al Poli y in the 'Thirties: A Reappraisal,� Ameri an E onomi

Review, 46(5), 857�879.

Chandler, L. V. (1971): Ameri an Monetary Poli y, 1928�1941. Harper & Row, Publish-

ers, New York.

Chow, G. C., and A. Lin (1971): �Best Linear Unbiased Interpolation, Distribution, and

Extrapolation of Time Series by Related Series,� Review of E onomi s and Statisti s,

53(4), 372�375.

Christiano, L. J., M. Ei henbaum, and C. L. Evans (1999): �Monetary Poli y Sho ks:

What Have We Learned and to What End?,� in Handbook of Ma roe onomi s, ed. by J. B.

Taylor, and M. Woodford, vol. 1A, pp. 65�148. Elsevier S ien e, Amsterdam.

Co hrane, J. H. (1999): �A Fri tionless View of U.S. In�ation,� in NBER Ma roe onomi s

Annual 1998, ed. by B. S. Bernanke, and J. J. Rotemberg, vol. 13, pp. 323�384. MIT Press,

Cambridge, MA.

(2001): �Long Term Debt and Optimal Poli y in the Fis al Theory of the Pri e

Level,� E onometri a, 69(1), 69�116.

Coddington, A. (1976): �Keynesian E onomi s: The Sear h for First Prin iples,� Journal

of E onomi Literature, 14(2), 1258�1273.

66

Ja obson, Leeper, & Preston: 1933

Cole, H. L., and L. E. Ohanian (2004): �New Deal Poli ies and the Persisten e of the

Great Depression: A General Equilibrium Analysis,� Journal of Politi al E onomy, 112(4),

779�816.

Currie, L. B., H. D. White, and P. T. Ellsworth (2002): �Memorandum of January

1932,� History of Politi al E onomy, 34(3), 533�552.

Cushman, D. O., and T. Zha (1997): �Identifying Monetary Poli y in a Small Open

E onomy Under Flexible Ex hange Rates,� Journal of Monetary E onomi s, 39(August),

433�448.

Department of the Treasury (various): �Annual Report of the Se retary of the Trea-

sury on the State of Finan es,� Washington, D.C., various �s al years.

E les, M. (1933): �Investigation of E onomi Problems,� Hearings Before the Senate

Finan e Committee, U.S. Congress, February 13�28.

E les, M. S. (1951): Be koning Froniters. Alfred A. Knopf, New York.

Eggertsson, G., and M. Woodford (2003): �The Zero Bound on Interest Rates and

Optimal Monetary Poli y,� Brookings Papers on E onomi A tivity, 2003(1), 139�211.

Eggertsson, G. B. (2008): �Great Expe tations and the End of the Depression,� Ameri an

E onomi Review, 98(4), 1476�1516.

Ei hengreen, B. (1992): Golden Fetters: The Gold Standard and the Great Depression.

Oxford University Press, New York.

(2000): �Was There a Paradigm Shift in E onomi Poli y in the 1930s?,� in Storia

dell'e onomia mondiale (A History of the World E onomy), Volume 4: Between Expansion

and Re ession (1850-1930), ed. by V. Castronovo. Editori Laterza, Rome.

Federal Reserve Board (1943): �Banking and Monetary Statisti s, 1914�1941,� Wash-

ington, D.C., September.

Firestone, J. M. (1960): Federal Re eipts and Expenditures during Business Cy les, 1879�

1958. Prin eton University Press, Prin eton, N.J.

Fishba k, P. (2010): �US Monetary and Fis al Poli y in the 1930s,� Oxford Review of

E onomi Poli y, 26(3), 385�413.

Fisher, I. (1932): Booms and Depressions. Adelphi Company, New York.

(1933a): Letter from Fisher to FDR, April 30.

(1933b): �The Debt-De�ation Theory of Great Depressions,� E onometri a, 1(4),

337�357.

(1934): After Re�ation, What? Adelphi Company, New York.

67

Ja obson, Leeper, & Preston: 1933

Friedman, M., and A. J. S hwartz (1963): A Monetary History of the United States,

1867-1960. Prin eton University Press, Prin eton, NJ.

Garbade, K. D. (2012): Birth of a Market. MIT Press, Cambridge, MA.

Goodfriend, M. (1988): �Central Banking Under the Gold Standard,� vol. 29 of Carnegie-

Ro hester Conferen e Series on Publi Poli y, pp. 85�124. Amsterdam.

Gordon, R. A. (1952): Business Flu tuations. Harper & Brothers, New York.

Hall, G. J., and T. J. Sargent (2015): �A History of U.S. Debt Limits,� Manus ript,

New York University, De ember 8.

Hausman, J. K. (2013): �New Deal Poli ies and Re overy from the Great Depression,�

Ph.D. dissertation, University of California, Berkeley, Spring.

Irwin, D. A. (2012): �Gold Sterilization and the Re ession of 1937�1938,� Finan ial History

Review, 19(3), 249�267.

Jalil, A. J., and G. Rua (2017): �In�ation Expe tations in the U.S. in Fall 1933,� in

Resear h in E onomi History, ed. by S. Wol ott, and C. Hanes, vol. 33, pp. 139�169.

Emerald Group Publishing Limited.

Jaremski, M., and G. Mathy (2016): �How was the Quantitative Eas-

ing Program of the 1930s Unwound?,� Manus ript, Ameri an University,

http://www.ameri an.edu/ as/e onomi s/resear h/upload/2016-01.pdf.

Johnson, Jr., G. G. (1939): The Treasury and Monetary Poli y, 1933�1938. Russell &

Russell, New York.

Keynes, J. M. (1933): �From Keynes to Roosevelt: Our Re overy Plan Assayed,� New

York Times, De ember 31.

Klüh, U., and P. Stella (2018): �Believing in Monetary Madness: How Roosevelt

Changed In�ation Expe tations,� Manus ript, Stellar Consulting LLC.

Leeper, E. M. (1991): �Equilibria Under `A tive' and `Passive' Monetary and Fis al Poli-

ies,� Journal of Monetary E onomi s, 27(1), 129�147.

Leeper, E. M., and C. Leith (2017): �Understanding In�ation as a Joint Monetary-Fis al

Phenomenon,� in Handbook of Ma roe onomi s, ed. by J. B. Taylor, and H. Uhlig, vol. 2B,

pp. 2305�2416. Elsevier Press, Amsterdam.

Leeper, E. M., C. A. Sims, and T. Zha (1996): �What Does Monetary Poli y Do?,�

Brookings Papers on E onomi A tivity, 1996(2), 1�63.

Leeper, E. M., and T. B. Walker (2013): �Per eptions and Misper eptions of Fis al

In�ation,� in Fis al Poli y After the Finan ial Crisis, ed. by A. Alesina, and F. Giavazzi,

pp. 255�299. University of Chi ago Press, Chi ago.

68

Ja obson, Leeper, & Preston: 1933

Leeper, E. M., and X. Zhou (2013): �In�ation's Role in Optimal Monetary-Fis al Poli y,�

NBER Working Paper No. 19686, November.

Leu htenburg, W. E. (1963): Franklin D. Roosevelt and the New Deal: 1932�1940.

Harper & Row, New York.

Meltzer, A. H. (2003): A History of the Federal Reserve, Volume I: 1913�1951. University

of Chi ago Press, Chi ago.

Meyer, E. (1934): �The Federal Reserve System,� The Washington Post, February 17, p.

8.

Miller, A. C. (1936): �The Return to Gold,� Address Before the A ademy of Politi al

S ien e, New York City, April 2.

Offi er, L. H., and S. H. Williamson (2018): �The Annual Con-

sumer Pri e Index for the United States, 1774�Present,� MeasuringWorth,

http://www.measuringworth. om/us pi/.

Paris, J. D. (1938): Monetary Poli ies of the United States, 1932�1938. Columbia Univer-

sity Press, New York.

Pearson, F. A., W. I. Meyers, and A. R. Gans (1957): �Warren as Presidential Advi-

sor,� Farm E onomi s, 211(De ember), 5597�5676.

Peppers, L. C. (1973): �Full-Employment Surplus Analysis and Stru tural Change: The

1930s,� Explorations in E onomi History, 10(2), 197�210.

Rau hway, E. (2014): �Going O� Gold and the Basis for Bretton Woods,� University of

California, Berkeley, E onomi History Seminar, Mar h 31.

(2015): The Money Makers. Basi Books, New York.

Raynold, P., W. D. M Millin, and T. R. Beard (1991): �The Impa t of Federal

Government Expenditures in the 1930s,� Southern E onomi Journal, 58(1), 15�28.

Romer, C. D. (1992): �What Ended the Great Depression?,� Journal of E onomi History,

52(4), 757�784.

Roosevelt, F. D. (1932a): �Campaign Address on the Federal Budget,� Pittsburgh, PA,

O tober 19.

(1932b): �Radio Address,� Albany, NY, April 7.

(1933a): �Letter to Treasury Se retary Woodin,� President's O� ial File 230,

September 30.

(1933b): �Press Conferen e #6, Mar h 24, 1933,�

http://www.fdrlibrary.marist.edu/_resour es/images/p /p 0184.pdf.

69

Ja obson, Leeper, & Preston: 1933

(1933 ): �Wireless to the London Conferen e,� July 3, 1933. Online

by Gerhard Peters and John T. Woolley, The Ameri an Presiden y Proje t.

http://www.presiden y.u sb.edu/ws/?pid=14679.

(1936a): �A eptan e Spee h for the Renomination for the Presiden y,� Philadel-

phia, PA, June 27. Online by Gerhard Peters and John T. Woolley, The Ameri an Presi-

den y Proje t. http://www.presiden y.u sb.edu/ws/?pid=15314.

(1936b): �Address at Forbes Field,� Pittsburgh, PA, O tober 1.

(1936 ): �Annual Budget Message,� President Roosevelt's Annual Budget Message

delivered to Congress.

Rosen, E. A. (2005): Roosevelt, the Great Depression, and the E onomi s of Re overy.

University of Virginia Press, Charlottesville, VA.

Sargent, T. J., and N. Walla e (1981): �Some Unpleasant Monetarist Arithmeti ,�

Federal Reserve Bank of Minneapolis Quarterly Review, 5(Fall), 1�17.

Sims, C. A. (1994): �A Simple Model for Study of the Determination of the Pri e Level and

the Intera tion of Monetary and Fis al Poli y,� E onomi Theory, 4(3), 381�399.

(2013): �Paper Money,� Ameri an E onomi Review, 103(2), 563�584.

(2016): �Making Government Paper a Bad Investment,� Presentation slides at

Be ker-Friedman Institute Conferen e �Next Steps for the Fis al Theory of the Pri e

Level,� April 1.

Sims, C. A., and T. Zha (1998): �Bayesian Methods for Dynami Multivariate Models,�

International E onomi Review, 39(November), 949�968.

(1999): �Error Bands for Impulse Responses,� E onometri a, 67(5), 1113�1155.

Stein, H. (1996): The Fis al Revolution in Ameri a. AEI Press, Washington, D.C., se ond

revised edn.

Steindl, F. G. (2004): Understanding E onomi Re overy in the 1930s: Endogenous Prop-

agation in the Great Depression. University of Mi higan Press, Ann Arbor.

Studenski, P., and H. E. Krooss (1952): Finan ial History of the United States. Beard

Books, Washington, D.C.

Sumner, S. (2001): �Roosevelt, Warren, and the Gold-Buying Program of 1933,� in Resear h

in E onomi History, ed. by S. Wol ott, and C. Hanes, vol. 20, pp. 135�172. Emerald Group

Publishing Limited.

Temin, P., and B. A. Wigmore (1990): �The End of One Big De�ation,� Explorations in

E onomi History, 27(4), 483�502.

70

Ja obson, Leeper, & Preston: 1933

Wheelo k, D. C. (1991): The Strategy and Consisten y of Federal Reserve Monetary

Poli y, 1924�1933. Cambridge University Press, New York.

Wi ker, E. (1966): Federal Reserve Monetary Poli y 1917�1933. Random House, New

York.

Wi ker, E. R. (1965): �Federal Reserve Monetary Poli y, 1922�33: A Reinterpretation,�

Journal of Politi al E onomy, 73(4), 325�343.

Woodford, M. (2001): �Fis al Requirements for Pri e Stability,� Journal of Money, Credit,

and Banking, 33(3), 669�728.

Zha, T. (1999): �Blo k Re ursion and Stru tural Ve tor Autoregressions,� Journal of E ono-

metri s, 90, 265�289.

71